Search NASA⌕ Search

SEARCH · Search NASA

Results for “dynamical model”

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 1,135 records · Page 63

Importance of filter’s microstructure in dynamic filtration modeling of gasoline particulate filters (GPFs): Inhomogeneous porosity and pore size distribution

The state-of-the-art multiscale modeling of GPFs including channel scale, wall scale, and pore scale is described. The microstructures of two GPFs were experimentally characterized. The pore size distributions of the GPFs were determined by mercury porosimetry. The porosity was measured by X-ray computed tomography (CT) and found to be inhomogeneous across the substrate wall. The significance of pore size distribution with respect to filtration performance was analyzed. The predictions of filtration efficiency were improved by including the pore size distribution in the filtration model. A dynamic heterogeneous multiscale filtration (HMF) model was utilized to simulate particulate filtration on a single channel particulate filter with realistic particulate emissions from a spark-ignition direct-injection (SIDI) gasoline engine. The dynamic evolution of filter’s microstructure and macroscopic filtration characteristics including mass- and number-based filtration efficiencies and pressure drop were predicted and discussed. The microstructure of the GPF substrate including inhomogeneous porosity and pore size distribution is found to significantly influence local particulate deposition inside the substrate and macroscopic filtration performance and is recommended to be resolved in the filtration model to simulate and evaluate the filtration performance of GPFs.

Environmental Molecular Sciences Laboratory↗

Learning compact physics‐aware delayed photocurrent models using dynamic mode decomposition

Abstract Radiation‐induced photocurrent in semiconductor devices can be simulated using complex physics‐based models, which are accurate, but computationally expensive. This presents a challenge for implementing device characteristics in high‐level circuit simulations where it is computationally infeasible to evaluate detailed models for multiple individual circuit elements. In this work we demonstrate a procedure for learning compact delayed photocurrent models that are efficient enough to implement in large‐scale circuit simulations, but remain faithful to the underlying physics. Our approach utilizes dynamic mode decomposition (DMD), a system identification technique for learning reduced‐order discrete‐time dynamical systems from time series data based on singular value decomposition. To obtain physics‐aware device models, we simulate the excess carrier density induced by radiation pulses by solving numerically the ambipolar diffusion equation, then use the simulated internal state as training data for the DMD algorithm. Our results show that the significantly reduced‐order delayed photocurrent models obtained via this method accurately approximate the dynamics of the internal excess carrier density—which can be used to calculate the induced current at the device boundaries—while remaining compact enough to incorporate into larger circuit simulations.

Hanson, Joshua↗

Plant-specific Model and Data Analysis using Dynamic Security Modeling and Simulation

The requirements for U.S. nuclear power plants to maintain a large on-site physical security force contribute to their high operational costs. The cost of maintaining the current physical security posture is approximately 10% of the overall operation and maintenance budget for commercial nuclear power plants. The goal of the Light Water Reactor Sustainability (LWRS) program’s physical security pathway is to develop tools, methods, and technologies and provide the technical basis for an optimized physical security posture. The conservatisms built into current security postures may be analyzed and minimized in order to reduce security costs while still ensuring adequate security and operational safety. The research performed at Idaho National Laboratory within LWRS program’s physical security pathway has successfully developed a dynamic force-on-force modeling framework using various computer simulation tools and integrating them with the dynamic assessment Event Modeling Risk Assessment using Linked Diagrams (EMRALD) tool. This document provides an update on the progress in applying a dynamic computational framework that links results from a commercially available force-on-force simulation tool, a commercially available thermal-hydraulic tool, and EMRALD to an operating commercial nuclear power plant. This report is only a summary of the progress and does not contain specific modeling results as those contain sensitive security information. This process of including plant procedures and multiple analysis results is being called Modeling and Analysis for Safety Security using Dynamic EMRALD Framework or MASS-DEF. Previous reports described how a user could integrate their plant-specific force-on-force models with the dynamic simulation tool EMRALD, model operator actions, integrate with probabilistic risk assessment tools, such as CAFTA (Computer Aided Fault Tree Analysis System) or SAPHIRE (Systems Analysis Programs for Hands-on Integrated Reliability Evaluations), and with thermal-hydraulic tools, such as RELAP-5. Previous reports applied various combinations of available simulations codes with EMRALD using generic plant models to demonstrate how to perform the analysis. This report documents the results of applying the dynamic computational framework to an actual nuclear facility using their security scenarios and timelines. This report does not contain any plant's sensitive information and/or Safeguards Information. The purpose of this study was to verify that results achieved using generic models are similar to actual plant results and to refine our guidance on the use of the framework. This assessment enables further analysis, such as what-if scenarios and staff-reduction evaluation, thereby optimizing physical security at plants.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Analysis of an unswept propfan blade with a semiempirical dynamic stall model

The time history response of a propfan wind tunnel model with dynamic stall is studied analytically. The response obtained from the analysis is compared with available experimental data. The governing equations of motion are formulated in terms of blade normal modes which are calculated using the COSMIC-NASTRAN computer code. The response analysis considered the blade plunging and pitching motions. The lift, drag and moment coefficients for angles of attack below the static stall angle are obtained from a quasi-steady theory. For angles above static stall angles, a semiempirical dynamic stall model based on a correction to angle of attack is used to obtain lift, drag and moment coefficients. Using these coefficients, the aerodynamic forces are calculated at a selected number of strips, and integrated to obtain the total generalized forces. The combined momentum-blade element theory is used to calculate the induced velocity. The semiempirical stall model predicted a limit cycle oscillation near the setting angle at which large vibratory stresses were observed in an experiment. The predicted mode and frequency of oscillation also agreed with those measured in the experiment near the setting angle.

Reddy, T. S. R.↗

Applying Particle Swarm Optimization and Extended Kalman Filtering to Model Kaplan Generation Dynamics for Hydropower Systems

Variable renewable generation is increasing the need for hydropower plants to provide fast and flexible grid support, which places new demands on plant-level dynamic models used for monitoring, control, and operational decision-making. This need is especially important for hydroelectric systems, where turbine and generator dynamics are strongly coupled, nonlinear, and time-varying, making accurate real-time representation difficult. To address this problem, this paper develops a digital twin (DT) framework for a synchronous generator–Kaplan turbine system using an explicit separation of slow turbine dynamics and fast generator dynamics. The turbine subsystem is represented by a six-coefficient model, whose parameters are identified offline using particle swarm optimization, while the generator subsystem is updated online through an extended Kalman filter for real-time state and parameter estimation. These models are integrated within a closed-loop simulation that includes a proportional–integral–derivative–double-derivative governor and excitation system, allowing the DT to track plant behavior under realistic operating conditions. Unlike prior studies that treat turbine and generator modeling separately or rely mainly on simulated inputs, the proposed framework is validated using real operational data from a hydropower plant. Results show that the DT reproduces terminal voltage, active power, and reactive power with a normalized root mean square error of approximately 5%. This hybrid offline–online formulation constitutes the main contribution of the work, providing an adaptive and practically deployable DT for hydropower systems with direct relevance to control improvement, performance monitoring, and grid-support applications under high renewable penetration.

13 HYDRO ENERGY↗

X-Ray Emitting GHz-Peaked Spectrum Galaxies: Testing a Dynamical-Radiative Model with Broad-Band Spectra

In a dynamical-radiative model we recently developed to describe the physics of compact, GHz-Peaked-Spectrum (GPS) sources, the relativistic jets propagate across the inner, kpc-sized region of the host galaxy, while the electron population of the expanding lobes evolves and emits synchrotron and inverse-Compton (IC) radiation. Interstellar-medium gas clouds engulfed by the expanding lobes, and photoionized by the active nucleus, are responsible for the radio spectral turnover through free-free absorption (FFA) of the synchrotron photons. The model provides a description of the evolution of the GPS spectral energy distribution (SED) with the source expansion, predicting significant and complex high-energy emission, from the X-ray to the {gamma}-ray frequency domain. Here, we test this model with the broad-band SEDs of a sample of eleven X-ray emitting GPS galaxies with Compact-Symmetric-Object (CSO) morphology, and show that: (i) the shape of the radio continuum at frequencies lower than the spectral turnover is indeed well accounted for by the FFA mechanism; (ii) the observed X-ray spectra can be interpreted as non-thermal radiation produced via IC scattering of the local radiation fields off the lobe particles, providing a viable alternative to the thermal, accretion-disk dominated scenario. We also show that the relation between the hydrogen column densities derived from the X-ray (N{sub H}) and radio (N{sub HI}) data of the sources is suggestive of a positive correlation, which, if confirmed by future observations, would provide further support to our scenario of high-energy emitting lobes.

08 HYDROGEN↗

On incorporating damping and gravity effects in models of structural dynamics of the SCOLE configuration

The damping for structural dynamic models of flexible spacecraft is usually ignored and then added after modal frequencies and mode shapes are calculated. It is common practice to assume the same damping ratio for all modes, although it is known that damping due to bending and that due to torsion are sometimes ignored. Two methods of including damping in the modeling process from its onset are examined. First, the partial derivative equations of motion are analyzed for a pinned-pinned beam with damping. The end conditions are altered to handle bodies with mass and inertia for the Spacecraft Control Laboratory Experiment (SCOLE) configuration. Second, a massless beam approximation is used for the modes with low frequencies, and a clamped-clamped system is used to approximate the modes for arbitrarily high frequency. The model is then modified to include gravity effects and is compared with experimental results.

Taylor, Larry↗

External nozzle flap dynamic load measurements on F-15 S/MTD model

Dynamic pressure loads were obtained on 1/12 scale models of the F-15B production aircraft and the F-15 S/MTD experimental aircraft with rectangular nozzles and canards. Flight Mach numbers from 0.51 to 1.20 were studied for aircraft angles of attack from 0 to 10 deg and nozzle pressure ratios from 1.00 to 5.09. The results show that dynamic levels are lower in the internozzle region of twin rectangular nozzles than are levels found with twin axisymmetric nozzles. At other locations, the levels associated with both geometries are of the same order of magnitude when normalized by aircraft dynamic Q. At Mach number of 0.51, the loads spectrum is dominated by plume shock noise processes for both geometries. Above Mach 0.51, this mechanism is associated with either vortex bursting from a forward location or turbulent boundary layer separation over the nozzle external flaps. At supersonic speeds both geometries show significantly decreased load levels.

Seiner, John M.↗

The Potosi Reservoir Model 2013c, Property Modeling Update

As part of a larger project co-funded by the United States Department of Energy (US DOE) to evaluate the potential of formations within the Cambro-Ordovician strata above the Mt. Simon as potential targets for carbon sequestration in the Illinois and Michigan Basins, the Illinois Clean Coal Institute (ICCI) requested Schlumberger to evaluate the potential injectivity and carbon dioxide (CO2) plume size of the Cambrian Potosi Formation. The evaluation of this formation was accomplished using wireline data, core data, pressure data, and seismic data from this project as well as two other separately funded projects: the US DOE-funded Illinois Basin–Decatur Project (IBDP) being conducted by the Midwest Geological Sequestration Consortium (MGSC) in Macon County, Illinois, and the Illinois Industrial Carbon Capture and Sequestration (ICCS) project funded through the American Recovery and Reinvestment Act. In 2010, technical performance evaluations on the Cambrian Potosi Formation were performed through reservoir modeling. The data included formation tops from mud logs, well logs from the Verification Well #1 (VW1) and the Injection Well (CCS1), structural and stratigraphic formation from three dimensional (3D) seismic data, and field data from several waste water injection wells for Potosi Formation. The intention was for 2.2 million tons per annum (2 million tonnes per annum [MTPA]) of CO2 to be injected for 20 years. In the Task Error! Reference source not found., the 2010 Potosi heterogeneous model (referred to as the "Potosi Dynamic Model 2010") was re-run using a new injection scenario of 3.5 million tons per annum (3.2 MTPA) for 30 years. The extent of the Potosi Dynamic Model 2010, however, appeared too small for the new injection target. The models size was insufficient to accommodate the evolution of the plume. The new model, Potosi Dynamic Model 2013a, was built by extending the Potosi Dynamic Model 2010 grid to 30 by 30 mi (48 by 48 km), while preserving all property modeling workflows and layering. This model was retained as the base case. In the preceding Task [1], the Potosi reservoir model was updated to take into account the new data from the Verification Well #2 (VW2) which was drilled in 2012. The porosity and permeability modeling was revised to take into account the log data from the new well. Revisions of the 2010 modeling assumptions were also done on relative permeability, capillary pressures, formation water salinity, and the maximum allowable well bottomhole pressure. Dynamic simulations were run using the injection target of 3.5 million tons per annum (3.2 MTPA) for 30 years. This dynamic model was named Potosi Dynamic Model 2013b. In this Task, a new property modeling workflow was applied, where seismic inversion data guided the porosity mapping and geobody extraction. The static reservoir model was fully guided by PorosityCube interpretations and derivations coupled with petrophysical logs from three wells. The two main assumptions are: porosity features in the PorosityCube that correlate with lost circulation zones represent vugular zones, and that these vugular zones are laterally continuous. Extrapolation was done carefully to populate the vugular facies and their corresponding properties outside the seismic footprint up to the boundary of the 30 by 30 mi (48 by 48 km) model. Dynamic simulations were also run using the injection target of 3.5 million tons per annum (3.2 MTPA) for 30 years. This new dynamic model was named Potosi Dynamic Model 2013c. Reservoir simulation with the latest model gives a cumulative injection of 43 million tons (39 MT) in 30 years with a single well, which corresponds to 40% of the injection target. The injection rate is approx. 3.2 MTPA in the first six months as the well is injecting into the surrounding vugs, and declines rapidly to 1.8 million tons per annum (1.6 MTPA) in year 3 once the surrounding vugs are full and the CO2 start to reach the matrix. After, the injection rate declines gradually to 1.2 million tons per annum (1.1 MTPA) in year 18 and stays constant. This implies that a minimum of three (3) wells could be required in the Potosi to reach the injection target. The injectivity evaluated in this Task was higher compared to the preceding Task, since the current facies modeling (guided by the porosity map from the seismic inversion) indicated a higher density of vugs within the vugular zones. 5 As the CO2 follows the paths where vugs interconnection exists, a reasonably large and irregular plume extent was created. After 30 years of injection, the plume extends 13.7 mi (22 km) in E-W and 9.7 mi (16 km) in N-S directions. After injection finishes, the plume continues to migrate laterally, mainly driven by the remaining pressure gradient. After 60 years post-injection, the plume extends 14.2 mi (22.8 km) in E-W and 10 mi (16 km) in N-S directions, and remains constant as the remaining pressure gradient has become very low. Should the targeted cumulative injection of 106 million tons (96 MT) be achieved; a much larger plume extent could be expected. The increase of reservoir pressure at the end of injection is approximately 1,200 psia (8,274 kPa) around the injector and gradually decreases away from the well. The reservoir pressure increase is less than 10 psia (69 kPa) beyond 14 mi (23 km) away from injector. Should the targeted cumulative injection of 106 million tons (96 MT) be achieved; a much larger areal pressure increase could be expected. The reservoir pressure declines rapidly during the first 30 years post injection and the initial reservoir pressure is nearly restored after 100 years post-injection. The present evaluation is mainly associated with uncertainty on the vugs permeability and interconnectivity. The use of porosity mapping from seismic inversion might have reduced the uncertainty on the lateral vugs body distributions. However, major uncertainties on the Potosi vugs permeability remains. Therefore, injection test and pressure interference test among the wells could be considered to evaluate the local vugs permeability, extent, and interconnectivity. Facies modeling within the Potosi has yet to be thoroughly addressed. The carbonates during the time of deposition are believed to be regionally extensive. However, it may be worth delineating the reservoir with other regional wells or modern day analogues to understand the extent of the Potosi. More specifically, the model could incorporate lateral changes or trends if deemed necessary to represent facies transition. Data acquisitions to characterize the fracture pressure gradient, the formation water properties, the relative permeability, and the capillary pressure could also be considered in order to allow a more rigorous evaluation of the Potosi storage performance. A simulation using several injectors could also be considered to determine the required number of wells and appropriate spacing to achieve the injection target while taking into account the pressure interference.

Adushita, Yasmin↗

A multiscale coupling method for the modeling of dynamics of solids with application to brittle cracks

We present a multiscale model for numerical simulations of dynamics of crystalline solids. The method combines the continuum nonlinear elasto-dynamics model, which models the stress waves and physical loading conditions, and molecular dynamics model, which provides the nonlinear constitutive relation and resolves the atomic structures near local defects. The coupling of the two models is achieved based on a general framework for multiscale modeling - the heterogeneous multiscale method (HMM). We derive an explicit coupling condition at the atomistic/continuum interface. Application to the dynamics of brittle cracks under various loading conditions is presented as test examples.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

The Potosi Reservoir Model 2013

As a part of a larger project co-funded by the United States Department of Energy (US DOE) to evaluate the potential of formations within the Cambro-Ordovician strata above the Mt. Simon as potential targets for carbon sequestration in the Illinois and Michigan Basins, the Illinois Clean Coal Institute (ICCI) requested Schlumberger to evaluate the potential injectivity and carbon dioxide (CO2) plume size of the Cambrian Potosi Formation. The evaluation of this formation was accomplished using wireline data, core data, pressure data, and seismic data from the US DOE-funded Illinois Basin–Decatur Project (IBDP) being conducted by the Midwest Geological Sequestration Consortium in Macon County, Illinois. In 2010, technical performance evaluations on the Cambrian Potosi Formation were performed through reservoir modeling. The data included formation tops from mud logs, well logs from the VW1 and the CCS1 wells, structural and stratigraphic formation from three dimensional (3D) seismic data, and field data from several waste water injection wells for Potosi Formation. Intention was for two million tons per annum (MTPA) of CO2 to be injected for 20 years. In the preceding, the 2010 Potosi heterogeneous model (referred to as the "Potosi Dynamic Model 2010" in this topical report) was re-run using a new injection scenario; 3.2 MTPA for 30 years. The extent of the Potosi Dynamic Model 2010, however, appeared too small for the new injection target. It was not sufficiently large enough to accommodate the evolution of the plume. The new model, Potosi Dynamic Model 2013a, was built by extending the Potosi Dynamic Model 2010 grid to 30 miles x 30 miles (48.3km x48.3km), while preserving all property modeling workflows and layering. This model was retained as the base case of Potosi Dynamic Model 2013a. The Potosi reservoir model was updated to take into account the new data from the verification well VW2 which was drilled in 2012. The new porosity and permeability modeling was performed to take into account the log data from the new well. Revisions of the 2010 modeling assumptions were also done on relative permeability, capillary pressures, formation water salinity, and the maximum allowable well bottomhole pressure. Dynamic simulations were run using the injection target of 3.2 MTPA for 30 years. This new dynamic model was named Potosi Dynamic Model 2013b. Due to the major uncertainties on the vugs permeability, two models were built; the Pessimistic and Optimistic Cases. The Optimistic Case assumes vugs permeability of 9,000 mD, which is analog to the vugs permeability identified in the pressure fall off test of a waste water injector in the Tuscola site, approx. 40 miles (64.4km) away from the IBDP area. The Pessimistic Case assumes that the vugs permeability is equal to the log data, which does not take into account the permeability from secondary porosity. The probability of such case is deemed low and could be treated as the worst case scenario, since the contribution of secondary porosity to the permeability is neglected and the loss circulation events might correspond to a much higher permeability. It is considered important, however, to identify the range of possible reservoir performance since there are no rigorous data available for the vugs permeability. The Optimistic Case gives an average CO2 injection rate of 0.8 MTPA and cumulative injection of 26 MT in 30 years, which corresponds to 27% of the injection target. The injection rate is approx. 3.2 MTPA in the first year as the well is injecting into the surrounding vugs, and declines rapidly to 0.8 MTPA in year 4 once the surrounding vugs are full and the CO2 start to reach the matrix. This implies that according to this preliminary model, a minimum of four (4) wells could be required to achieve the injection target. This result is lower than the injectivity estimated in the Potosi Dynamic Model 2013a (43 MT in 30 years), since the permeability model applied in the Potosi Dynamic Model 2013b is more conservative. This revision was deemed necessary to treat the uncertainty in a more appropriate manner. As the CO2 follows the paths where vugs interconnection exists, a reasonably large and irregular plume extent was created. For the Optimistic Case, the plume extends 17 miles (27.4km) in E-W and 14 miles (22.5km) in N-S directions after 30 years. After injection is completed, the plume continues to migrate laterally, mainly driven by the remaining pressure gradient. After 100 years post injection, the plume extends 20 miles (32.2km) in E-W and 15.5 miles (24.9km) in N-S directions. Should the targeted cumulative injection of 96 MT be achieved; a much larger plume extent could be expected. For the Optimistic Case, the increase of reservoir pressure at the end of injection is approximately 1200 psia (8,274 kPa) around the injector and gradually decreases away from the well. The reservoir pressure increase is less than 30 psia (206.8 kPa) beyond 14 miles (22.5km) away from injector. Should the targeted cumulative injection of 96 MT be achieved; a much larger areal pressure increase could be expected. The initial reservoir pressure is nearly restored after approximately 100 years post injection. The presence of matrix slows down the pressure dissipations. The Pessimistic Case gives an average CO2 injection rate of 0.2 MTPA and cumulative injection of 7 MT in 30 years, which corresponds to 7% of the injection target. This implies that in the worst case scenario, a minimum of sixteen (16) wells could be required to achieve the injection target. The present evaluation is mainly associated with uncertainty on the vugs permeability, distribution, and interconnectivity. The different results indicated by the Optimistic and Pessimistic Cases signify the importance of vugs permeability characterization. Therefore, injection test and pressure interference test among the wells could be considered to evaluate the local vugs permeability, extent, and interconnectivity. Porosity mapping derived from the seismic inversion could also be used in the succeeding task to characterize the lateral porosity distribution within the reservoir. With or without seismic inversion porosity mapping, it is worth exploring whether increased lateral heterogeneity plays a significant role in Potosi injectivity. Investigations on vugular, dolomitic outcrops suggest that there may be significantly greater lateral heterogeneity than what has been modeled here. Facies modeling within the Potosi has yet to be thoroughly addressed. The carbonates during the time of deposition are believed to be regionally extensive. However, it may be worth delineating the reservoir with other regional wells or modern day analogues to understand the extent of the Potosi. More specifically, the model could incorporate lateral changes or trends if deemed necessary to represent facies transition. Data acquisitions to characterize the fracture pressure gradient, the formation water properties, the relative permeability, and the capillary pressure could also be considered in order to allow a more rigorous evaluation of the Potosi storage performance. A simulation using several injectors could also be considered to determine the required number of wells to achieve the injection target while taking into account the pressure interference.

Adushita, Yasmin↗

Chapter 15: Using System Dynamics to Model Industry's Developmental Response to Energy Policy

In this chapter we explore the potential development of the biofuels industry using the Biomass Scenario Model (BSM), a system dynamics model developed at the National Renewable Energy Laboratory through the support of the U.S. Department of Energy. The BSM is designed to analyze the implications of policy on the development of the supply chain for biofuels in the United States. It explicitly represents the behavior of decision makers such as farmers, investors, fueling station owners, and consumers. We analyze several illustrative case studies that explore a range of policies and discuss how incentives interact with individual parts of the supply chain as well as the industry as a whole. The BSM represents specific incentives that are intended to approximate policy in the form of selected laws and regulations. Through characterizing the decision making behaviors of economic actors within the supply chain that critically influence the adoption rate of new biofuels production technologies and demonstrating synergies among policies, we find that incentives with coordinated impacts on each major element of the supply chain catalyze net effects of decision maker behavior such that the combined incentives are greater than the summed effects of individual incentives in isolation.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗

Moving beyond post hoc explainable artificial intelligence: a perspective paper on lessons learned from dynamical climate modeling

AI models are criticized as being black boxes, potentially subjecting climate science to greater uncertainty. Explainable artificial intelligence (XAI) has been proposed to probe AI models and increase trust. In this review and perspective paper, we suggest that, in addition to using XAI methods, AI researchers in climate science can learn from past successes in the development of physics-based dynamical climate models. Dynamical models are complex but have gained trust because their successes and failures can sometimes be attributed to specific components or sub-models, such as when model bias is explained by pointing to a particular parameterization. We propose three types of understanding as a basis to evaluate trust in dynamical and AI models alike: (1) instrumental understanding, which is obtained when a model has passed a functional test; (2) statistical understanding, obtained when researchers can make sense of the modeling results using statistical techniques to identify input–output relationships; and (3) component-level understanding, which refers to modelers' ability to point to specific model components or parts in the model architecture as the culprit for erratic model behaviors or as the crucial reason why the model functions well. We demonstrate how component-level understanding has been sought and achieved via climate model intercomparison projects over the past several decades. Such component-level understanding routinely leads to model improvements and may also serve as a template for thinking about AI-driven climate science. Currently, XAI methods can help explain the behaviors of AI models by focusing on the mapping between input and output, thereby increasing the statistical understanding of AI models. Yet, to further increase our understanding of AI models, we will have to build AI models that have interpretable components amenable to component-level understanding. We give recent examples from the AI climate science literature to highlight some recent, albeit limited, successes in achieving component-level understanding and thereby explaining model behavior. The merit of such interpretable AI models is that they serve as a stronger basis for trust in climate modeling and, by extension, downstream uses of climate model data.

54 ENVIRONMENTAL SCIENCES↗

Method of discrete modeling and its application to estimation of TF30 engine variables

A method of discrete modeling is presented that effectively isolates steady state model accuracy from dynamic model accuracy. The steady state model may be generated from the engine design equations with any desired degree of accuracy. The dynamic model is generated by applying a step disturbance of a manipulated variable to an open loop engine simulation. The sampled response of the variable is combined with the steady state model's response to form a set of weighting factors. These weighting factors are then used to weight past values of the manipulated variable, thus forming the dynamic model. The method is used to estimate various TF30-P-3 engine variables. A dynamic trim function is developed to compensate for the dynamic nonlinearities of the variables as well as for inaccuracies in dynamic definition. The trim function is shown to be realted to the square root of the sum of the squares of the weighting factors obtained at various engine operating conditions. Finally, the estimation of variables without dynamic modeling is discussed.

Arpasi, D. J.↗

Short-lead seasonal precipitation forecast in northeastern Brazil using an ensemble of artificial neural networks

This study assesses the deterministic and probabilistic forecasting skill of a 1-month-lead ensemble of Artificial Neural Networks (EANN) based on low-frequency climate oscillation indices. The predictand is the February-April (FMA) rainfall in the Brazilian state of Ceará, which is a prominent subject in climate forecasting studies due to its high seasonal predictability. Additionally, the study proposes combining the EANN with dynamical models into a hybrid multi-model ensemble (MME). The forecast verification is carried out through a leave-one-out cross-validation based on 40 years of data. The EANN forecasting skill is compared with traditional statistical models and the dynamical models that compose Ceará’s operational seasonal forecasting system. A spatial comparison showed that the EANN was among the models with the smallest Root Mean Squared Error (RMSE) and Ranked Probability Score (RPS) in most regions. Moreover, the analysis of the area-aggregated reliability showed that the EANN is better calibrated than the individual dynamical models and has better resolution than Multinomial Logistic Regression for above-normal (AN) and below-normal (BN) categories. It is also shown that combining the EANN and dynamical models into a hybrid MME reduces the overconfidence of the extreme categories observed in a dynamically-based MME, improving the reliability of the forecasting system.

54 ENVIRONMENTAL SCIENCES↗