Search NASA⌕ Search

SEARCH · Search NASA

Results for “comparative modeling”

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 109 records · Page 6
All sourcesNASA NTRSDOE OSTI
DOE OSTI2023

Prediction and Validation of Flow Properties in Porous Lattice Structures

High-porosity metal foams have been extensively studied as an attractive candidate for efficient and compact heat exchanger design. With the advancements in additive manufacturing, such foams can be manufactured with controlled topology to yield highly tailorable mechanical and transport properties. In this study, a lattice Boltzmann method (LBM)-based pore-scale model is implemented to simulate the fluid flow in additively manufactured (AM) metal foams with unit cell topologies of Cube, Face Diagonal (FD)-Cube, Tetrakaidecahedron (TKD), and Octet lattices. The pressure gradient versus average velocity profiles predicted by the LBM model were validated against in-house measurements on the AM lattice samples with the same unit cell topologies. Based on the simulation results, a novel hybrid model is proposed to accurately predict the volume averaged flow properties (permeability and inertial coefficients) of the four structures. Specifically, the linear LBM (neglecting inertial forces) is first implemented to obtain the intrinsic permeability, and then the standard LBM is applied to obtain the inertial coefficient. Convenient correlations for those flow properties as a function of porosity and fiber diameter are constructed. The effects of the AM print qualities on the flow properties are also discussed. The advantages of the hybrid model compared to the polynomial fitting approach for determining flow properties are discussed and compared quantitatively. The hybrid model and presented results are valuable for flow and thermal transport evaluation when designing new metal foams for specific applications and with different materials and topologies. Finally, the presented correlations based on pore-scale simulations can also be conveniently used in volume-averaged models to predict the macroscale flow behavior in such complex structures.

42 ENGINEERING↗
DOE OSTI2020

EBS Task Force: Task 9/FEBEX Modeling Final Report: Thermo-Hydrological Modeling with PFLOTRAN

This report outlines Sandia National Laboratories modeling studies applied to Stage 1 and Stage 2 of the Full-scale Engineered Barriers Experiment in Crystalline Host Rock (FEBEX) in situ test for the SKB EBS Task Force Task 9. The FEBEX test was a full-scale test conducted over ~18 years at the Grimsel, Switzerland Underground Research Laboratory (URL) managed by NAGRA. It involved emplacing simulated waste packages, in the form of welded cylindrical heaters, inside a tunnel in crystalline granitic rock and surrounded by a bentonite barrier and cement plug. Sensors emplaced within the bentonite monitored the wetting-up, heating, and drying out of the bentonite barrier, and the large resulting data set provides an excellent opportunity for validation of multiphysics Thermal-Hydrological (TH), Thermal-Hydrologic-Chemical (THC), and Thermal-Hydrological-Mechanical (THM) modeling approaches for underground nuclear waste storage and the performance of engineered bentonite barriers. The present status of the EBS Task Force is finalizing Task 9, which follows years of modeling studies of the FEBEX test, by many notable modeling teams (Gens et al., 2009; Sanchez et al. 2010; 2012; Samper et al., 2018). These modeling studies generally use two-dimensional axisymmetric meshes, ignoring threedimensional effects, gravity and asymmetric wetting and dry out of the bentonite engineered barrier. This study investigates these effects with use of the PFLOTRAN THC code with massively parallel computational methods in modeling FEBEX Stage 1 and Stage 2 results. The PFLOTRAN numerical code is an open source, state-of-the-art, massively parallel subsurface flow and reactive transport code operating in a high-performance computing environment (Hammond et al., 2014). Section 2 describes the applied partial differential equations describing mass, momentum and energy balance used in this study, considerations derived by assuming phase equilibrium between gas and liquid phases, constitutive equations for granite, cement plug, and bentonite domains, and specific approaches for use inthe PFLOTRAN code. Section 3 describes the geometry, meshing, and model set-up. Section 4 describes modeling results, Section 5 compares modeling results to field testing data, and Section 6 gives conclusions. The Appendix provides detailed information required by the EBSTask Force for final reporting.

42 ENGINEERING↗
DOE OSTI2024

Simplified inelastic constitutive models for ASME Section III, Division 5 design by inelastic analysis

This report describes the development of simplified, universal constitutive model that captures the high temperature monotonic and cyclic behavior of a range of commonly-used high temperature materials. The goal of the work is to provide a simple, universal constitutive model to replace the current bespoke models for Grade 91, 316H, and Alloy 617 included in Nonmandatory Appendix HBB-Z of the ASME Boiler & Pressure Vessel Code, and to extend this model to cover Alloy 800H. We initiated this work in response to feedback from reactor vendors and other Code users requesting simplified models, compared to the current models, that are easier to implement and use in commercial finite element analysis software. This report describes the completion of this effort by developing a model to correct the defects in standard model forms presently used for high temperature material modeling, described in past work, developing and implementing new numerical methods to train this model against test data, and then actually training the model for the four materials. The report provides a complete mathematical description of the model along with the tabulated material coefficients for the four materials. The final step will be to formulate an ASME Code change to introduce the new models into the Code.

36 MATERIALS SCIENCE↗
NASA NTRS1975

A grid of model atmospheres for metal-deficient giant stars. I

A grid of flux-constant model atmospheres for stars with effective temperatures between 3750 and 6000 K, log g between 0.75 and 3.0, and (A/H) between -3.0 and 0.0 has been constructed. The line absorption is approximated by opacity distribution functions. Metal lines and molecular lines, including those from the infrared bands of CO and CN, are taken into account. The variation of the structure of the models with metal abundance and microturbulence parameter is found to be quite regular and not very drastic. The surface cooling produced by CO is important for all the models with a maximum temperature of 5000 K, while CN mainly causes a backwarming effect but is not very important for solar CNO abundances. The effects of convection, estimated by using the mixing-length approximation, are important only for the coldest models. The models compare very well with models from overlapping regions of other grids. A solar model consistent with the grid models is found to agree satisfactorily with empirical solar model atmospheres. The careful use of scaled solar models for stars with (A/H) approximately equal to zero is justified.

Gustafsson, B.↗
NASA NTRS1990

Determination of the ocean circulation using Geosat altimetry

A spherical harmonic model of the sea surface topography complete to degree and order 10 and a model of the earth's geopotential field complete to degree and order 50 have been obtained in a simultaneous solution using Geosat altimeter data and tracking data from 14 different satellites. The sea surface topography model compares well with oceanographic models computed using hydrographic data and ship drift data. Currently, errors in the estimated gravity field model limit the determination of the spherical harmonic coefficients of the general ocean circulation to degrees 10 and lower, corresponding to a minimum wavelength of 4000 km. Error analysis indicates that the correlation between the geoid and the sea surface topography model is less than 0.2, indicating good separation of the geoid and the sea surface topography at wavelengths of 4000 km or longer. Estimates of the scale factor for the significant wave height (H1/3), which is used to compute the electromagnetic bias correction and the bias for the Geosat altimeter, are obtained. The estimate of the H1/3 correction is 3.6 + or - 1.5 percent, and the height bias estimate is zero.

Nerem, R. S.↗
NASA NTRS1992

A model to compare performance of space and ground network support of low-Earth orbiters

This article compares the downlink performance in a gross average sense between space and ground network support of low-Earth orbiters. The purpose is to assess what the demand for DSN support of future small, low-cost missions might be, if data storage for spacecraft becomes reliable enough and small enough to support the storage requirements needed to enable support only a fraction of the time. It is shown that the link advantage of the DSN over space reception in an average sense is enormous for low-Earth orbiters. The much shorter distances needed to communicate with the ground network more than make up for the speedup in data rate needed to compensate for the short contact times with the DSN that low-Earth orbiters have. The result is that more and more requests for DSN-only support of low-Earth orbiters can be expected.

Posner, E. C.↗
NASA NTRS1993

Recent solar extreme ultraviolet irradiance observations and modeling: A review

For more than 90 years, solar extreme ultraviolet (EUV) irradiance modeling has progressed from empirical blackbody radiation formulations, through fudge factors, to typically measured irradiances and reference spectra was well as time-dependent empirical models representing continua and line emissions. A summary of recent EUV measurements by five rockets and three satellites during the 1980s is presented along with the major modeling efforts. The most significant reference spectra are reviewed and threee independently derived empirical models are described. These include Hinteregger's 1981 SERF1, Nusinov's 1984 two-component, and Tobiska's 1990/1991/SERF2/EUV91 flux models. They each provide daily full-disk broad spectrum flux values from 2 to 105 nm at 1 AU. All the models depend to one degree or another on the long time series of the Atmosphere Explorer E (AE-E) EUV database. Each model uses ground- and/or space-based proxies to create emissions from solar atmospheric regions. Future challenges in EUV modeling are summarized including the basic requirements of models, the task of incorporating new observations and theory into the models, the task of comparing models with solar-terrestrial data sets, and long-term goals and modeling objectives. By the late 1990s, empirical models will potentially be improved through the use of proposed solar EUV irradiance measurements and images at selected wavelengths that will greatly enhance modeling and predictive capabilities.

Tobiska, W. Kent↗
NASA NTRS2003

A Global Aerosol Model Forecast for the ACE-Asia Field Experiment

We present the results of aerosol forecast during the Aerosol Characterization Experiment (ACE-Asia) field experiment in spring 2001, using the Georgia Tech/Goddard Global Ozone Chemistry Aerosol Radiation and Transport (GOCART) model and the meteorological forecast fields from the Goddard Earth Observing System Data Assimilation System (GEOS DAS). The aerosol model forecast provides direct information on aerosol optical thickness and concentrations, enabling effective flight planning, while feedbacks from measurements constantly evaluate the model, making successful model improvements. We verify the model forecast skill by comparing model predicted total aerosol extinction, dust, sulfate, and SO2 concentrations with those quantities measured by the C-130 aircraft during the ACE-Asia intensive operation period. The GEOS DAS meteorological forecast system shows excellent skills in predicting winds, relative humidity, and temperature for the ACE-Asia experiment area as well as for each individual flight, with skill scores usually above 0.7. The model is also skillful in forecast of pollution aerosols, with most scores above 0.5. The model correctly predicted the dust outbreak events and their trans-Pacific transport, but it constantly missed the high dust concentrations observed in the boundary layer. We attribute this missing dust source to the desertification regions in the Inner Mongolia Province in China, which have developed in recent years but were not included in the model during forecasting. After incorporating the desertification sources, the model is able to reproduce the observed high dust concentrations at low altitudes over the Yellow Sea. Two key elements for a successful aerosol model forecast are correct source locations that determine where the emissions take place, and realistic forecast winds and convection that determine where the aerosols are transported. We demonstrate that our global model can not only account for the large-scale intercontinental transport, but also produce the small-scale spatial and temporal variations that are adequate for aircraft measurements planning.

Chin, Mian↗
NASA NTRS1990

A dynamic growth model of vegetative soya bean plants: model structure and behaviour under varying root temperature and nitrogen concentration

A differential equation model of vegetative growth of the soya bean plant (Glycine max (L.) Merrill cv. Ransom') was developed to account for plant growth in a phytotron system under variation of root temperature and nitrogen concentration in nutrient solution. The model was tested by comparing model outputs with data from four different experiments. Model predictions agreed fairly well with measured plant performance over a wide range of root temperatures and over a range of nitrogen concentrations in nutrient solution between 0.5 and 10.0 mmol NO3- in the phytotron environment. Sensitivity analyses revealed that the model was most sensitive to changes in parameters relating to carbohydrate concentration in the plant and nitrogen uptake rate.

Non-NASA Center↗
NASA NTRS2011

GEM-CEDAR Challenge: Comparing Ionospheric Models with Poynting Flux from DMSP Observations

As part to the GEM-CEDAR challenge we are extending the model-data comparisons to electrodynamic in-situ measurements in low-Earth orbit. We use DMSP observations of electric and magnetic fields to compute Poynting Flux values along the satellite track in high latitudes including the auroral zones and the polar cap. Models of the ionosphere that include electrodynamic parameters have been run for five events selected for the GEM-CEDAR modeling challenge for which DMSP data are available for comparison. Combined with a magnetic field model we use the modeled electric fields to compute Poynting Flux and Joule Dissipation values from outputs of CTIPe, TIE-GCM, the ionospheric electrodynamics solvers of the SWMF, LFM and OpenGGCM magnetosphere-ionosphere coupled models, and the Weimer electric field model. The online metrics analysis tool at the Community Coordinated Modeling Center (CCMC) has been updated to handle the analysis of separate short segments of available data (high-latitude sections of the satellite orbit) with model outputs to analyze how well auroral patterns are being reproduced by the models. We present initial results from the new analysis tool in terms of model yields (ratio of the difference between maximum and minimum values of model results to the observation), timing/location errors of local maxima in the inbound and outbound auroral crossings as well as cross-correlations for individual passes. We collect the information for many DMSP passes and present an analysis for model performance during quiet and geomagnetically disturbed time periods using half-orbit integrated values as well.

Rastaetter, Lutz↗
NASA NTRSDate not supplied

3D Material Response of the MSL Heatshield Using NuSil-Coated PICA

The Mars Science Laboratory (MSL) was protected during its atmospheric entry by an instrumented heatshield that used NASA's Phenolic Impregnated Carbon Ablator (PICA) material [1]. PICA is a lightweight carbon fiber/polymeric resin material that offers outstanding performance for protecting probes during planetary entry. Data from the Mars Entry Descent and Landing Instrument (MEDLI) suite on MSL offers unique in-flight validation data for models of material response and atmospheric entry. MEDLI recorded, among other things, time-resolved in-depth temperature data of PICA using thermocouple sensors assembled in the MEDLI Integrated Sensor Plugs (MISP) [2]. A space-grade silicone-based coating commercially known as NuSil CV-1144-0 [3] was applied to the entire MSL heatshield, including the MEDLI plugs, to mitigate the spread of dust from PICA. Modeling the thermal response of PICA-NuSil (PICA-N) system is still an open challenge. Ground testing of PICA-N models exhibited surface temperature jumps of the order of 150 K due to oxide scale formation and sub-sequent NuSil burn-off. It is therefore critical to include a validated model for the material response of the coating in engineering codes. A test campaign has been conducted at the NASA’s Langley HyMETS [4] facility to screen the response of PICA-N and gather detailed data on its behavior [5]. A first model of PICA-N thermal response has been developed using the Hy-METS experiments [6]. The objective of this work is to analyze the material response of the latest PICA-N model compared to the engineering model used to simulate the entry of MSL. The environment and material response around the MSL aeroshell during Mars atmospheric entry is simulated using a collection of tools. The Direct Simulation Monte Carlo SPARTA code [7] is used in the rarefied regime, the Data Parallel Line Relaxation (DPLR) code [8] is used in the continuum regime and radiative heating conditions are provided by the Nonequilibrium air radiation (NEQAIR) code [9] to estimate the environmental conditions. The thermal response inside the material is computed using the Porous material Analysis Toolbox based on Open-FOAM (PATO) [10,11,12]. Thermodynamic and chemistry properties are estimated using the Mutation++ library [13]. The approach implemented in PATO as a first cut PICA-N thermal response model is outlined in Figure 1. While the recession is less than the coating thickness, the Surface mass and energy balance Boundary Condition (SBC) uses the NuSil B’ tables. Once the recession removes the coating, the usual PICA B’ tables are used for the SBC. The B’ tables are computed using an equilibrium solver implemented in Mutation++, given the temperature, pressure, blowing rate, composition of the pyrolysis and environment gases, and the condensed species at the surface. Preliminary results of the 3D material response of the MSL heat-shield at the peak heating (80 sec after Entry Interface) are shown in Figure 2. Current NASA’s mission to Mars, Mars 2020, used the spare heatshield of MSL for thermal protection during entry, descent, and landing. In preparation for Mars 2020 post-flight analysis, the PATO high-fidelity material response capability was benchmarked against flight data from MEDLI. This effort represents an important milestone toward the development of validated predictive capabilities for designing thermal protection systems for planetary probes. This bench-marking is awaiting the final release of the MEDLI-2 data.

Aerospace↗
NASA NTRS2010

Non-Targeted Effects and the Dose Response for Heavy Ion Tumorigenesis

BACKGROUND: There is no human epidemiology data available to estimate the heavy ion cancer risks experienced by astronauts in space. Studies of tumor induction in mice are a necessary step to estimate risks to astronauts. Previous experimental data can be better utilized to model dose response for heavy ion tumorigenesis and plan future low dose studies. DOSE RESPONSE MODELS: The Harderian Gland data of Alpen et al.[1-3] was re-analyzed [4] using non-linear least square regression. The data set measured the induction of Harderian gland tumors in mice by high-energy protons, helium, neon, iron, niobium and lanthanum with LET s ranging from 0.4 to 950 keV/micron. We were able to strengthen the individual ion models by combining data for all ions into a model that relates both radiation dose and LET for the ion to tumor prevalence. We compared models based on Targeted Effects (TE) to one motivated by Non-targeted Effects (NTE) that included a bystander term that increased tumor induction at low doses non-linearly. When comparing fitted models to the experimental data, we considered the adjusted R2, the Akaike Information Criteria (AIC), and the Bayesian Information Criteria (BIC) to test for Goodness of fit.In the adjusted R2test, the model with the highest R2values provides a better fit to the available data. In the AIC and BIC tests, the model with the smaller values of the summary value provides the better fit. The non-linear NTE models fit the combined data better than the TE models that are linear at low doses. We evaluated the differences in the relative biological effectiveness (RBE) and found the NTE model provides a higher RBE at low dose compared to the TE model. POWER ANALYSIS: The final NTE model estimates were used to simulate example data to consider the design of new experiments to detect NTE at low dose for validation. Power and sample sizes were calculated for a variety of radiation qualities including some not considered in the Harderian Gland data set and with different background tumor incidences. We considered different experimental designs with varying number of doses and varying low doses dependant on the LET of the radiation. The optimal design to detect a NTE for an individual ion had 4 doses equally spaced below a maximal dose where bending due to cell sterilization was < 2%. For example at 100 keV/micron we would irradiate at 0.03 Gy, 0.065 Gy, 0.13 Gy, and 0.26 Gy and require 850 mice including a control dose for a sensitivity to detect NTE with 80% power. Sample sizes could be improved by combining ions similar to the methods used with the Harderian Gland data.

Chappelli, Lori J.↗
NASA NTRS2002

Rationalizing Burned Carbon with Carbon Monoxide Exported from South America

We present several estimates cross-checking the fluxes of carbon to the atmosphere from burning, comparing models that are based on simple land-surface parameterizations and atmospheric transport dynamics. Both estimates made by NASA Ames and USP modeling techniques are quite high compared to some detailed satellite/land-use studies of emissions. The flux of carbon liberated to the atmosphere via biomass burning is important for several reasons. This flux is a fundamental statistic for the parameterization of the large-scale flux of gases controlling the reactive greenhouse gases methane and ozone. Similarly, it is central to the estimation of the translocation of nitrogen and pyrodenitrification in the tropics. Thirdly, CO2 emitted from rainforest clearing contributes directly to carbon lost from the rainforest system as it contributes to greenhouse gas forcing. While CO2 from pasturage, agriculture, etc, is considered to be reabsorbed seasonally, and so "off budget" for the carbon cycle, it must also be accounted. CO2 anomalies related to daily weather and interannual climatic variation are strong enough to perturb our scientific perception of long-term carbon storage trends. We compare fluxes deduced from land-use statistics (originally, W.M. Hao) and from satellite hot pixels (A. Setzer) with atmospheric fluxes determined by the mesoscale/continental scale models RAMS and MM5, and point to some new work with highly resolved global models (the NASA Data Assimilation Office's GEOS4). Our simulations are tied to events, so that measured tracers like CO tie the models directly to the burning and meteorology of a specific period. We point out a particular sensitivity in estimates based on CO, and indicate how analysis of CO2 along with other biomass-burning tracers may lead to an improved multi-species estimator of carbon burned.

Chatfield, R.↗
DOE OSTI2023

Physics-constrained Gaussian process model for prediction of hydrodynamic interactions between wave energy converters in an array

To improve the efficiency of wave farms and achieve maximum power generation, the layout of wave energy converters (WECs) in an array needs to be carefully designed so that the hydrodynamic interactions can be positively exploited. For this, the hydrodynamic characteristics of the WEC array in different layouts need to be calculated. However, such calculations using numerical models usually entail significant computational cost, especially for large arrays of WECs. To address the computational challenge, a physics-constrained Gaussian process (GP) model is proposed to replace the original expensive numerical model and predict the hydrodynamic characteristics of the WECs for any array layout. By exploring the relationship between the WEC array (i.e., the input) and different hydrodynamic characteristics (i.e., the output), here we summarize a set of physical constraints/features, including invariance, symmetry, and additivity. This prior knowledge about the input-output relationship is then directly embedded in the constructed GP model through the design of physics-constrained kernels. In particular, a double-sum invariant kernel is first developed to incorporate the invariance and symmetry features, and then an additive kernel is developed to incorporate the additive feature of the problem. The invariant kernel and the additive kernel are then integrated to construct the physics-constrained GP model. Compared to the standard GP model, the proposed physics-constrained GP models require less training data to achieve the desired accuracy in predicting the hydrodynamic characteristics and are also less vulnerable to the curse of dimensionality (i.e., good scalability for large arrays) due to the use of an additive kernel. The efficiency, accuracy, and scalability of the proposed approach are demonstrated through an application to predict the hydrodynamic characteristics for WEC arrays of different sizes and layouts.

16 TIDAL AND WAVE POWER↗
DOE OSTI2025

Physics-Informed Graph Neural Networks for Collaborative Dynamic Reconfiguration and Voltage Regulation in Unbalanced Distribution Systems

Network reconfiguration has long been employed as a strategic approach to minimize power distribution system losses and effectively regulate voltage levels. Tap-changing voltage regulators are also critical for controlling bus voltages, especially in accommodating the increasing integration of distributed energy resources (DERs) with intermittent outputs. This paper introduces novel methodologies to address the challenges of dynamic reconfiguration and optimal tap setting in unbalanced three-phase distribution systems. We propose an approximated mixed-integer quadratically constrained program (MIQCP) to model dynamic reconfiguration, along with a pioneering formulation for voltage regulator (VR) tap-setting based on Special Ordered Set type 1 (SOS1). To mitigate computational complexity, we propose a physics-informed spatial-temporal graph convolutional network (STGCN) with an integrated link classifier. The proposed approach enables efficient solution generation by fixing specific variables in the MIQCP instance and solving the simplified sub-MIP using an MIP solver. Numerical studies demonstrate the superior prediction accuracy of our STGCN model compared to baseline neural network models, resulting in reduced DER curtailment and voltage deviation with shorter computation time.

dynamic reconfiguration↗
DOE OSTI2025

Simulating High-Frequency Seismograms in Realistic Earth Models to Better Understand Source Discrimination Based on Differential Magnitudes ( M L− M c)

Discriminating low-yield underground nuclear explosions from small earthquakes is a key task in monitoring nuclear test ban treaties. P/S amplitude ratios have been an effective discriminant for moderate-sized events recorded at regional distances, but it is unclear if they are as effective in discriminating small seismic events recorded at local distances (<150 km). The difference between local magnitude (M L ) and coda duration magnitude (M c ) has been proposed as a new discriminant that may complement P/S amplitude ratios at local distances. Here, in this work, we calculate high-frequency (up to ∼4 Hz) synthetic seismograms at epicentral distances of 0–30 km in realistic models of the Salt Lake basin (Utah, United States) to better understand how variations in source type and depth affect M L −M c values. The Earth models incorporate simplified 1D and deterministic 3D structures, small-wavelength stochastic velocity perturbations, and surface topography. Coda waves are enhanced for the more complicated models compared to the base 1D model, but still underpredict observed durations by about a factor of two, which results in overprediction of amplitude to duration ratios (i.e., M L −M c values) for a near-surface explosion and a 7 km deep earthquake. For both source types, the predicted M L and M c values decrease as source depth increases, and M L −M c shows only minor variation with depth; however, M L −M c is on average ∼0.5 units smaller for explosions than earthquakes. This finding may imply that M L −M c has sensitivity to source type, in addition to being a depth discriminant, but more modeling is needed given the limitations of the current study. Future modeling should incorporate higher-frequency (≳5 Hz) simulations over a larger distance range (0–150 km), where M L and M c are commonly measured, while honoring low shear velocities (<300 m/s) near the surface and sampling a wider range of earthquake and explosion source mechanisms.

Hutchings, Sean J. [Univ. of Utah, Salt Lake City,↗
DOE OSTI2024

Teaching Freight Mode Choice Models New Tricks Using Interpretable Machine Learning Methods

Understanding and forecasting the intricate freight mode choice behavior under various industry, policy, and technology contexts is essential in freight planning and policymaking. Numerous models have been developed in prior studies to provide insights into freight mode selection, the majority of which use discrete choice models such as multinomial logit (MNL) models. However, logit models often rely on linear specifications of independent variables, despite potential nonlinear relationships in the data. Moreover, there often lacks a heuristic and efficient approach to identify such complex relationships to define the logit model specifications. To fill this gap, we developed an MNL model for freight mode choice using the insights from state-of-the- art machine learning (ML) models. ML models can capture the nonlinear nature of the complex decision-making process, and recent advances in 'explainable AI' have greatly improved their interpretability. The interpretable ML methods help enhance the performance of MNL models and advance knowledge of freight mode choice. Specifically, the influential factors and their relationship with individual modes are identified using SHapley Additive exPlanations (SHAP) to improve the MNL's performance. The workflow is demonstrated in a case study of Austin, Texas, and the SHAP results reveal multiple nonlinear relationships predicted by ML models. Incorporating those relationships into MNL model specifications improves the interpretability and accuracy of the MNL model compared to a conventional MNL model. Findings from this study can be used to guide freight planning and inform policymakers and practitioners on how key factors affect freight decision-making.

ADVANCED PROPULSION SYSTEMS,MATHEMATICS AND COMPUT↗