Stochastic Generation Research Articles

Improving power system performance by integrating TCSC and SVC devices in the presence of stochastic renewable energy generators

Improving power system performance by integrating TCSC and SVC devices in the presence of stochastic renewable energy generators

Stochastic Scenario Generation Methods for Uncertainty in Wind and Photovoltaic Power Outputs: A Comprehensive Review

Stochastic Scenario Generation Methods for Uncertainty in Wind and Photovoltaic Power Outputs: A Comprehensive Review

Fast spatial simulation of extreme high-resolution radar precipitation data using integrated nested Laplace approximations

Abstract Aiming to deliver improved precipitation simulations for hydrological impact assessment studies, we develop a methodology for modelling and simulating high-dimensional spatial precipitation extremes, focusing on both their marginal distributions and tail dependence structures. Tail dependence is crucial for assessing the consequences of extreme precipitation events, yet most stochastic weather generators do not attempt to capture this property. The spatial distribution of precipitation occurrences is modelled with four competing models, while the spatial distribution of nonzero extreme precipitation intensities are modelled with a latent Gaussian version of the spatial conditional extremes model. Nonzero precipitation marginal distributions are modelled using latent Gaussian models with gamma and generalized Pareto likelihoods. Fast inference is achieved using integrated nested Laplace approximations. We model and simulate spatial precipitation extremes in Central Norway, using 13 years of hourly radar data with a spatial resolution of 1×1km2, over an area of size 6,461km2, to describe the behaviour of extreme precipitation over a small drainage area. Inference on this high-dimensional data set is achieved within hours, and the simulations capture the main trends of the observed precipitation well.

A hybrid statistical–dynamical framework for compound coastal flooding analysis

Abstract Compound coastal flooding due to astronomic, atmospheric, oceanographic, and hydrologic forcings poses severe threats to coastal communities. While physics-driven approaches are able to dynamically simulate temporally and spatially varying compound flooding generated by multiple partially correlated drivers, computational burdens limit their capability to explore the full range of conditions that contribute to compound coastal hazards. Data-driven statistical approaches address some of these computational challenges, however they are also unable to explore all possible forcing combinations due to short observational records, and projections are typically limited to a few locations. This study proposes a hybrid statistical-dynamical framework for compound coastal flooding analysis that integrates a stochastic generator of compound flooding drivers, a hydrodynamic model, and a machine learning-based surrogate model. The framework is demonstrated in San Francisco (SF) Bay over the past 500 years with high accuracy and computational efficiency. The stochastic generator of compound flooding drivers is developed by coupling a sea surface temperature (SST) reconstruction model with a climate emulator, weather generator, and hydrologic model. Using reconstructed SSTs as input, the generator of compound flooding drivers is employed to simulate time series of the forcing factors contributing to compound flooding in SF Bay. A process-based hydrodynamic model is built to predict total water levels (TWLs) varying in time and space throughout SF Bay based on the stochastically generated drivers. A machine learning-based surrogate model is then developed from a relatively small library of hydrodynamic model simulations to efficiently predict water levels (WLs) for compound flooding analysis under the full range of stochastic drivers. This study contributes a hybrid statistical-dynamical framework to better understand the spatial distribution and temporal evolution of compound coastal flooding, along with the relative contributions of the forcing drivers in complex nearshore, estuarine, and river environments for centennial timescales under past, present, and future climates.

Predictive uncertainty in state-estimation drives active sensing.

Animals use active sensing movements to shape the spatiotemporal characteristics of sensory signals to better perceive their environment under varying conditions. However, the underlying mechanisms governing the generation of active sensing movements are not known. To address this, we investigated the role of active sensing movements in the refuge tracking behavior ofEigenmannia virescens, a species of weakly electric fish. These fish track the longitudinal movements of a refuge in which they hide by swimming back and forth in a single linear dimension. During refuge tracking,Eigenmanniaexhibits stereotyped whole-body oscillations when the quality of the sensory signals degrades. We developed a closed-loop feedback control model to examine the role of these ancillary movements on the task performance. Our modeling suggests that fish may use active sensing to minimize predictive uncertainty in state estimation during refuge tracking. The proposed model generates simulated fish trajectories that are statistically indistinguishable from that of the actual fish, unlike the open-loop noise generator and stochastic resonance generator models in the literature. These findings reveal the significance of closed-loop control in active sensing behavior, offering new insights into the underlying mechanisms of dynamic sensory modulation.

Stochastic force generation in an isometric binary mechanical system.

Accurate models of muscle contraction are necessary for understanding muscle performance and the molecular modifications that enhance it (e.g., therapeutics, posttranslational modifications, etc.). As a thermal system containing millions of randomly fluctuating atoms that on the thermal scale of a muscle fiber generate unidirectional force and power output, muscle mechanics are constrained by the laws of thermodynamics. According to a thermodynamic muscle model, muscle's power stroke occurs with the shortening of an entropic spring consisting of an ensemble of force-generating myosin motor switches, each induced by actin binding and gated by inorganic phosphate release. This model differs fundamentally from conventional molecular power stroke models that assign springs to myosin motors in that it is physically impossible to describe an entropic spring in terms of the springs of its molecular constituents. A simple two-state thermodynamic model (a binary mechanical system) accurately accounts for muscle force-velocity relationships, force transients following rapid mechanical and chemical perturbations, and a thermodynamic work loop. Because this model transforms our understanding of muscle contraction, it must continue to be tested. Here, we show that a simple stochastic kinetic simulation of isometric muscle force predicts four phases of a force-generating loop that bifurcates between periodic and stochastic beating through mechanisms framed by two thermodynamic equations. We compare these model predictions with experimental data including observations of spontaneous oscillatory contractions (SPOCs) in muscles and periodic force generation in small myosin ensembles.

Bayesian Belief Networks: Redefining wholesale electricity price modelling in high penetration non-firm renewable generation power systems

The transition of electricity generation from firm to non-firm renewable generation is driving changes in wholesale electricity price dynamics that are increasingly challenging to model due to the stochastic nature of wind and solar as the primary energy source. Robust pricing models are essential for optimising financial performance in liberalised electricity markets. The novelty of this paper is the application of Bayesian Belief Networks in the modelling of wholesale electricity price formation, specifically in power systems with a high penetration of non-firm renewable generation. This paper links the mathematical Bayesian representation to established statistical and computational approaches using a functional supply-side wholesale electricity market pricing model. In addition, the paper introduces a novel validation method employing volatility analysis to assess the case study's performance.The case study revealed that when the proportion of stochastic generation was ≥40 % penetration in a trading interval, the Bayesian Belief Network model's accuracy considerably outperformed the benchmark model with a Mean Absolute Scale Error of 1.43 compared to 1.63 in those intervals. Finally, the volatility analysis results challenged the prevailing notion of a positive causal relationship between non-firm renewable penetration and price volatility.

Conditional neural field latent diffusion model for generating spatiotemporal turbulence

Eddy-resolving turbulence simulations are essential for understanding and controlling complex unsteady fluid dynamics, with significant implications for engineering and scientific applications. Traditional numerical methods, such as direct numerical simulations (DNS) and large eddy simulations (LES), provide high accuracy but face severe computational limitations, restricting their use in high-Reynolds number or real-time scenarios. Recent advances in deep learning-based surrogate models offer a promising alternative by providing efficient, data-driven approximations. However, these models often rely on deterministic frameworks, which struggle to capture the chaotic and stochastic nature of turbulence, especially under varying physical conditions and complex, irregular geometries. Here, we introduce the Conditional Neural Field Latent Diffusion (CoNFiLD) model, a generative learning framework for efficient high-fidelity stochastic generation of spatiotemporal turbulent flows in complex, three-dimensional domains. CoNFiLD synergistically integrates conditional neural field encoding with latent diffusion processes, enabling memory-efficient and robust generation of turbulence under diverse conditions. Leveraging Bayesian conditional sampling, CoNFiLD flexibly adapts to various turbulence generation scenarios without retraining. This capability supports applications such as zero-shot full-field flow reconstruction from sparse sensor data, super-resolution generation, and spatiotemporal data restoration. Extensive numerical experiments demonstrate CoNFiLD’s capability to accurately generate inhomogeneous, anisotropic turbulent flows within complex domains. These findings underscore CoNFiLD’s potential as a versatile, computationally efficient tool for real-time unsteady turbulence simulation, paving the way for advancements in digital twin technology for fluid dynamics. By enabling rapid, adaptive high-fidelity simulations, CoNFiLD can bridge the gap between physical and virtual systems, allowing real-time monitoring, predictive analysis, and optimization of complex fluid processes.

Tailoring Electrolyzer Catalyst Layer Morphology Via Pore Network Modelling

Polymer electrolyte membrane (PEM) water electrolysis is a key technology to produce clean, high purity hydrogen, which can serve as an alternative to carbon-based fuels. However, PEM electrolyzer uptake has been sluggish due to high catalyst costs, which can contribute to up to 47% of total stack costs [1]. Thus, to lower stack costs, the morphology of the catalyst layer must be optimized towards lower precious metal loadings. However, catalyst layer morphology optimization remains a challenge, as the mechanisms linking catalyst morphology to mass and charge transport remain ambiguous in the literature [2], [3]. While the morphology of porous transport layer (PTL) has been extensively explored via stochastic material generation and pore network modelling techniques [4], such an analysis must also be extended to the catalyst layer to design efficient, low-cost materials.In this work, we introduce a method to replicate the pore structure of commercially available PEM electrolyzer catalyst layers using stochastic generation techniques. Our method employs a custom divider to combine regions of varying pore sizes and effectively replicate the range of catalyst layer pore sizes observed in commercial materials. Moreover, we demonstrate that the pore size distribution of stochastically generated materials is statistically indistinguishable from imaged catalyst samples. Pore network modelling techniques were further applied to stochastically generated morphologies to evaluate electrical and mass transport properties. We not only reveal that the transport properties of our stochastically generated materials fall within experimentally measured ranges, but also reveal how tailoring pore size distributions can enhance these properties. The findings from our analysis can be used to identify desirable catalyst morphology targets for manufacturing and enhance the accuracy of electrolyzer transport models via the estimation of catalyst layer transport properties.[1] A. Mayyas, M. Ruth, et al., Manufacturing Cost Analysis for Proton Exchange Membrane Water Electrolyzers, United States (2019).[2] Z. Taie, X. Peng, et al., ACS Appl. Mater. Interfaces, 12, 47, 52701–52712 (2020).[3] J. Lopata, Z. Kang, et al., J. Electrochem. Soc., 167. 064507 (2020).[4] J. K. Lee, C. H. Lee, and A. Bazylak, J. Power Sources, 437, 226910 (2019).

A Raster-Based Multi-Objective Spatial Optimization Framework for Offshore Wind Farm Site-Prospecting

Siting an offshore wind project is considered a complex planning problem with multiple interrelated objectives and constraints. Hence, compactness and contiguity are indispensable properties in spatial modeling for Renewable Energy Sources (RES) planning processes. The proposed methodology demonstrates the development of a raster-based spatial optimization model for future Offshore Wind Farm (OWF) multi-objective site-prospecting in terms of the simulated Annual Energy Production (AEP), Wind Power Variability (WPV) and the Depth Profile (DP) towards an integer mathematical programming approach. Geographic Information Systems (GIS), statistical modeling, and spatial optimization techniques are fused as a unified framework that allows exploring rigorously and systematically multiple alternatives for OWF planning. The stochastic generation scheme uses a Generalized Hurst-Kolmogorov (GHK) process embedded in a Symmetric-Moving-Average (SMA) model, which is used for the simulation of a wind process, as extracted from the UERRA (MESCAN-SURFEX) reanalysis data. The generated AEP and WPV, along with the bathymetry raster surfaces, are then transferred into the multi-objective spatial optimization algorithm via the Gurobi optimizer. Using a weighted spatial optimization approach, considering and guaranteeing compactness and continuity of the optimal solutions, the final optimal areas (clusters) are extracted for the North and Central Aegean Sea. The optimal OWF clusters, show increased AEP and minimum WPV, particularly across offshore areas from the North-East Aegean (around Lemnos Island) to the Central Aegean Sea (Cyclades Islands). All areas have a Hurst parameter in the range of 0.55–0.63, indicating greater long-term positive autocorrelation in specific areas of the North Aegean Sea.

Investigation of the Effect of Integrated Offset, GPS, and InSAR Data in the Stochastic Source Modeling of the 2002 Denali Earthquake

This study investigates the effect of geological field measurement (offset), global positioning system (GPS), and interferometric synthetic aperture radar (InSAR) data on the estimation of the co-seismic earthquake displacements of the 2002 Denali earthquake. The analysis is conducted using stochastic source modeling. Uncertainties associated with each dataset limit their effectiveness in source model selection and raise questions about the adequate number of datasets and their type for reliable source estimation. To address these questions, stochastic source models with heterogeneous earthquake slip distributions are synthesized using the von Kármán wavenumber spectrum and statistical scaling relationships. The surface displacements of the generated stochastic sources are obtained using the Okada method. The surface displacements are compared with the available datasets (i.e., offset, GPS, and InSAR) individually and in an integrated form. The results indicate that the performance of stochastic source generation can be significantly improved in the case of using GPS data and in the integrated case. Overall, based on the case study of the 2002 Denali earthquake, the combined use of all available datasets increases the robustness of the stochastic source modeling method in characterizing surface displacement. However, GPS data contribute more than InSAR and offset data in producing reliable source models.

Stochastic Generation of Peak Ground Accelerations Based on Single Seismic Event Data for Safety Assessment of Structures

The Korean Peninsula, characterized by low-to-moderate seismicity, faces a shortage of strong ground motion records, posing challenges for the seismic safety assessment of critical infrastructures. Given the rarity of large-magnitude earthquakes, generating a variety of earthquakes with rational values of Peak Ground Acceleration (PGA) is essential for robust seismic fragility and risk analysis. To address this, a new stochastic approach is proposed to simulate artificial earthquakes at multiple source-to-site distances and derive the probability distribution of PGA based on recorded data from a single seismic event. Two key source parameters, seismic moment and corner frequency, are treated as random variables with a negative correlation, reflecting their uncertainties and dependence on source-to-site distance. The Monte Carlo simulation with copula sampling of the key source parameters generates Fourier spectra for artificial earthquakes, which are transformed into the time domain to yield PGA distributions at various distances. A comparison with recorded data shows that the proposed method effectively simulates ground motion intensities, with no statistically significant differences between the simulated results and recorded data (p>0.05). The present method of determining PGA distributions provides a robust framework to enhance seismic risk analysis for the safety assessment of structures.

Impact of stochastic vehicle generation on traffic microsimulation model output

One of the most important characteristics of microsimulation is the ability to model the temporal and spatial nature of traffic demand. Every traffic microsimulation has a vehicle generation model that determines how and when the driver–vehicle units are introduced in the simulation. While microsimulation use has become increasingly popular, it is unclear how the vehicle generation options affect the final result. The purpose of this paper is to examine the stochastic component of the vehicle generation model and how it may affect the simulation output. Three scenarios, including the passenger car equivalent (PCE) model of the Highway Capacity Manual (HCM-7), were used for assessing stochastic volumes and their impacts on performance measures. Results indicate a statistically significant impact of the variability of stochastic volumes on the estimation of performance measures at the breakdown flow phase and synchronized flow phase for interrupted and uninterrupted flow conditions, respectively. This finding highlights the importance of reporting vehicle generation as a calibration parameter, enabling others to replicate the experiments. When utilizing microsimulations for the assessment of roadway/system performance, it is crucial for users to have a thorough understanding of demand generation.

Multivariate stochastic generation of meteorological data for building simulation through interdependent meteorological processes

In recent years, the uncertainty of weather conditions and the impact of future climate change on building energy assessment has received increasing attention. As an important part of these studies, several types of methods for generating stochastic meteorological data have also been developed. Since solar radiation drops to zero at night, unlike the continuous 24-hour data for elements such as temperature and humidity, this has posed challenges for previous research to fully account for the simultaneity among multiple elements. Therefore, this study proposes a framework for meteorological data generation: First, perform multivariate time series modeling of meteorological data of air temperature, solar radiation and absolute humidity at 12:00 of each day of a typical year based on the S-vine copula method and simulating daily series data at 12:00 for 365 days. Then, based on the probability of change of each element evaluated from the historical meteorological observation data, the daily series data at 12:00 were expanded to 24 h, after which the yearly stochastic weather data were obtained. The analysis of 30 years of stochastic data generated by this method, compared with the original data, reveals that air temperature and solar radiation closely match the original distribution characteristics, except for a minor deviation in the absolute humidity’s kurtosis. Furthermore, the comparison of thermal load distributions for office buildings shows that the original data curve falls within the range of the generated data. This suggests that the generated data includes more information about uncertainty but still keeps the original data’s characteristics.

Methods to Evaluate Subcolumn Profiles Based on Two‐Point Diagnostics

AbstractIn atmospheric models, stochastic generation of subgrid‐scale profiles or “subcolumns” has been used for a variety of purposes. Such subcolumns can be generated from subgrid probability density functions (PDFs) at different vertical levels, when such PDFs are available. To do so, the generator needs to decide how strongly points should be correlated in the vertical, that is, how much the values should be overlapped. This is sometimes called “PDF overlap.” To assess vertical correlation in a simplified, observable setting, here the vertical correlation of vertical velocity in subcloud layers is examined. Doppler lidar is used to evaluate the vertical profiles of vertical velocity produced by a large‐eddy simulation (LES) model and the Subgrid Importance Latin Hypercube Sampler (SILHS) subcolumn generator. In order to diagnose unrealistic features in subcolumn profiles, various statistical diagnostics are examined here, including the bivariate PDF of vertical velocity at two separated points (i.e., altitudes), the two‐point velocity correlation, the integral correlation length, the PDF of two‐point velocity differences, and the skewness and kurtosis of two‐point velocity differences. The profiles produced by LES match lidar well, except that they are too smooth at small scales. The profiles produced by SILHS exhibit sharp jumps from updraft to downdraft that are not observed in the lidar data. To reduce the generation of these unrealistically sharp jumps, the SILHS sampling method is revised. The diagnostics confirm that the revised sampling method reduces the overprediction of sharp jumps.

A comparative study of different deep learning methods for time-series probabilistic residential load power forecasting

The widespread adoption of nonlinear power electronic devices in residential settings has significantly increased the stochasticity and uncertainty of power systems. The original load power data, characterized by numerous irregular, random, and probabilistic components, adversely impacts the predictive performance of deep learning techniques, particularly neural networks. To address this challenge, this paper proposes a time-series probabilistic load power prediction technique based on the mature neural network point prediction technique, i.e., decomposing the load power data into deterministic and stochastic components. The deterministic component is predicted using deep learning neural network technology, the stochastic component is fitted with Gaussian mixture distribution model and the parameters are fitted using great expectation algorithm, after which the stochastic component prediction data is obtained using the stochastic component generation method. Using a mature neural network point prediction technique, the study evaluates six different deep learning methods to forecast residential load power. By comparing the prediction errors of these methods, the optimal model is identified, leading to a substantial improvement in prediction accuracy.

Carbon-binder-domain porosity extraction through lithium-ion battery electrode impedance data

In the field of 3-D resolved computational modelling of Lithium-ion battery electrodes, the arrangement and properties of the Carbon-Binder-Domain (CBD) play a critical role in the ion and electron transport properties through their impact on the electrode tortuosity factor. However, until now, the CBD porosity value -its main descriptor in terms of transport properties and occupied volume- has been determined through educated guesses due to the lack of an experimental approach. Here, a novel methodology is reported for the determination of the CBD internal porosity through the combination of computational modelling and experimental electrochemical impedance spectroscopy (EIS). The methodology is based on the creation of a calibration curve that relates tortuosity factor with CBD porosity through digital stochastic generation of electrode microstructures and diffusivity characterization. The curve is then compared to the EIS experimental results and analyzed through a transmission line model, yielding a good estimation of the parameters. In this work, the usefulness and the identified limitation of this approach are demonstrated using three different formulations of LiNi0.3Mn0.3Co0.3O2 (NMC 111) cathodes. To the best of the authors’ knowledge, this is the first reported method for estimating CBD porosity.

Text-Driven Video Prediction

Current video generation models usually convert signals indicating appearance and motion received from inputs (e.g., image and text) or latent spaces (e.g., noise vectors) into consecutive frames, fulfilling a stochastic generation process for the uncertainty introduced by latent code sampling. However, this generation pattern lacks deterministic constraints for both appearance and motion, leading to uncontrollable and undesirable outcomes. To this end, we propose a new task called Text-driven Video Prediction (TVP). Taking the first frame and text caption as inputs, this task aims to synthesize the following frames. Specifically, appearance and motion components are provided by the image and caption separately. The key to addressing the TVP task depends on fully exploring the underlying motion information in text descriptions, thus facilitating plausible video generation. In fact, this task is intrinsically a cause-and-effect problem, as the text content directly influences the motion changes of frames. To investigate the capability of text in causal inference for progressive motion information, our TVP framework contains a Text Inference Module (TIM), producing step-wise embeddings to regulate motion inference for subsequent frames. In particular, a refinement mechanism incorporating global motion semantics guarantees coherent generation. Extensive experiments are conducted on Something-Something V2 and Single Moving MNIST datasets. Experimental results demonstrate that our model achieves better results over other baselines, verifying the effectiveness of the proposed framework.

Partially adaptive multistage stochastic programming

Multistage stochastic programming is a powerful tool allowing decision-makers to revise their decisions at each stage based on the realized uncertainty. However, organizations are not able to be fully flexible, as decisions cannot be revised too frequently in practice. Consequently, decision commitment becomes crucial to ensure that initially made decisions remain unchanged for a certain period of time. This paper introduces partially adaptive multistage stochastic programming, a new optimization paradigm that strikes an optimal balance between decision flexibility and commitment by determining the best stages to revise decisions depending on the allowed level of flexibility. We introduce a novel mathematical formulation and theoretical properties eliminating certain constraint sets. Furthermore, we develop a decomposition method that effectively handles mixed-integer partially adaptive multistage programs by adapting the integer L-shaped method and Benders decomposition. Computational experiments on stochastic lot-sizing and generation expansion planning problems show substantial advantages attained through optimal selections of revision times when flexibility is limited, while demonstrating computational efficiency attained by employing the proposed properties and solution methodology. By adhering to these optimal revision times, organizations can achieve performance levels comparable to fully flexible settings.

A theoretical study of a radiofrequency wave propagation through a realistic turbulent marine atmospheric boundary layer based on large eddy simulation

This study aims at modeling and investigating the impact of realistic turbulence inhomogeneity on radiowave propagation by utilizing large eddy simulations (LES). An up-to-date version of the X-LES method for generating realistic turbulent phase screens is first introduced. It combines atmospheric simulations with the classical Tatarski statistical modeling. This method naturally incorporates vertical turbulence inhomogeneity into phase screens at scales resolved by LES. It is then extended to sub-grid scales by weighting statistically generated phase variations with the vertical profile of the turbulent structure constant extracted from atmospheric data. This method is applied to replicate turbulence of a classical tropical marine atmospheric boundary layer. The impact of the generated medium on the propagation of a 10 GHz spherical wave emanating from a Gaussian aperture is analyzed through a statistical study of log-amplitude profiles performed for three different source altitudes. Results first show that contrary to a classical homogeneous turbulence modeling, log-amplitude profiles resulting from the propagation into inhomogeneous turbulence exhibit a statistical heterogeneity strongly dependent of source altitude. Furthermore, classical stochastic phase screen generation from a homogeneous Von-Kàrmàn Kolmogorov spectrum seems to give a statistically significant underestimation of the actual impact of turbulence compared to the X-LES method.