Search NASA⌕ Search

SEARCH · Search NASA

Results for “Surface process complexity”

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 415 records · Page 23

Impact of Small-Alkane Solvents on Polyolefin Hydrogenolysis over a Ruthenium Catalyst

Selective catalytic hydrogenolysis of polyolefins is a promising route to convert plastic waste into valuable liquid products, such as lubricants, waxes, and surfactants. However, the high viscosity of polymer melts imposes mass transfer limitations on this reaction. Solvents can mitigate these challenges, but their effects on reaction kinetics and product selectivity remain underexplored. Here, we systematically explore the effects of small n-alkanes and cycloalkanes on the hydrogenolysis of polyethylene and polypropylene over a Ru/TiO 2 catalyst. Using kinetic measurements and isotopic labeling, we show that n-octane at high mass fractions alters the mechanism from direct hydrogenation to solvent-mediated hydrogen transfer, reducing the rate of C–C bond cleavage. Longer alkanes further inhibit reactivity due to stronger surface binding. 1,4-Dimethylcyclohexane suppresses methane formation, favoring heavier products, while decalin likely forms surface-bound aromatics that poison the catalyst. Overall, alkane solvents modulate product selectivity and reduce the yield of methane byproduct, allowing for ∼35–40% selectivity to valuable C 20 -C 30 alkane products. This work highlights the complex impact of polymer–alkane mixtures on hydrogenolysis kinetics relevant to the design of commercial-scale plastic waste valorization processes.

Catalysts↗

Motility of Synthetic Cells from Engineered Lipids

Synthetic cells are artificial systems that resemble natural cells. Significant efforts have been made over the years to construct synthetic protocells that can mimic biological mechanisms and perform various complex processes. These include compartmentalization, metabolism, energy supply, communication, and gene reproduction. Cell motility is also of great importance, as nature uses elegant mechanisms for intracellular trafficking, immune response, and embryogenesis. In this review, we discuss the motility of synthetic cells made from lipid vesicles and relevant molecular mechanisms. Synthetic cell motion may be classified into surface-based or solution-based depending on whether it involves interactions with surfaces or movement in fluids. Collective migration behaviors have also been demonstrated. The swarm motion requires additional mechanisms for intercellular signaling and directional motility that enable communication and coordination among the synthetic vesicles. In addition, intracellular trafficking for molecular transport has been reconstituted in minimal cells with the help of DNA nanotechnology. These efforts demonstrate synthetic cells that can move, detect, respond, and interact. Finally, we envision that new developments in protocell motility will enhance our understanding of biological processes and be instrumental in bioengineering and therapeutic applications.

59 BASIC BIOLOGICAL SCIENCES↗

Flower-shaped 2D crystals grown in curved fluid vesicle membranes

Abstract The morphologies of two-dimensional (2D) crystals, nucleated, grown, and integrated within 2D elastic fluids, for instance in giant vesicle membranes, are dictated by an interplay of mechanics, permeability, and thermal contraction. Mitigation of solid strain drives the formation of crystals with vanishing Gaussian curvature (i.e., developable domain shapes) and, correspondingly, enhanced Gaussian curvature in the surrounding 2D fluid. However, upon cooling to grow the crystals, large vesicles sustain greater inflation and tension because their small area-to-volume ratio slows water permeation. As a result, more elaborate shapes, for instance, flowers with bendable but inextensible petals, form on large vesicles despite their more gradual curvature, while small vesicles harbor compact planar crystals. This size dependence runs counter to the known cumulative growth of strain energy of 2D colloidal crystals on rigid spherical templates. This interplay of intra-membrane mechanics and processing points to the scalable production of flexible molecular crystals of controllable complex shape.

36 MATERIALS SCIENCE↗

Atomistic Mechanism Underlying the Si ( 111 ) - ( 7 × 7 ) Surface Reconstruction Revealed by Artificial Neural-Network Potential

The 7×7 reconstruction of the Si(111) surface represents arguably the most fascinating surface reconstruction so far observed in nature. Yet, the atomistic mechanism underpinning its formation remains unclear after it was discovered sixty years ago. Experimentally, it is observed post priori so that analysis of its formation mechanism can only be carried out in analogy with archaeology. Theoretically, density-functional theory (DFT) correctly predicts the Si(111)-(7×7) ground state but is impractical to simulate its formation process; while empirical potentials failed to produce it as the ground state. Developing an artificial neural-network potential of DFT quality, we carried out accurate large-scale simulations to unravel the formation of the Si(111)-(7×7) surface. We reveal a possible step-mediated atom-pop rate-limiting process that triggers massive nonconserved atomic rearrangements, most remarkably, a critical process of collective vacancy diffusion that mediates a sequence of selective dimer, corner-hole, stacking-fault, and dimer-line pattern formation, to fulfill the 7×7 reconstruction. Our findings may not only solve the long-standing mystery of this famous surface reconstruction but they also illustrate the power of machine learning in studying complex structures.

36 MATERIALS SCIENCE↗

Impact of Additive Manufacturing on Internal Cooling Channels With Varying Diameters and Build Directions

Abstract The use of additive manufacturing (AM) processes, such as direct metal laser sintering, provides the design freedom required to incorporate complex cooling schemes in gas turbine components. Additively manufactured turbine components have a range of cooling feature sizes and, because of the inherent three-dimensionality, a wide range of build angles. Previous studies have shown that AM built directions influence internal channel surface roughness that, in turn, augment heat transfer and pressure loss. This study investigates the impact of AM on channel feature size and builds direction relative to tolerance, surface roughness, pressure losses, and convective cooling. Multiple AM coupons were built from Inconel 718 consisting of channels with different diameters and a variety of build directions. An experimental rig was used to measure pressure drop to calculate friction factor and was used to impose a constant surface temperature boundary condition to collect Nusselt number over a range of Reynolds numbers. Significant variations in surface roughness and geometric deviations from the design intent were observed for distinct build directions and channel sizes. These differences led to notable impacts in friction factor and Nusselt number augmentations, which were a strong function of build angle.

Engineering↗

Toward Hybrid Physics-Machine Learning to improve Land Surface Model predictions

A critical challenge for Land Surface Models (LSMs) is to simulate processes at the surface and the subsurface and their feedbacks to the atmosphere. Even using the same climate forcings, different LSMs predict different surface fluxes and soil moisture conditions due to differences in the formulations of individual processes, parameterizations, and representation of spatial heterogeneity. Ultimately, these differences contribute to the LSM prediction errors and uncertainty. This research seeks to address this challenge by coupling physics-based modeling with state-of-the-art machine learning (ML) techniques to describe complex physical and biogeochemical processes and narrow the gap between model predictions and observations.

58 GEOSCIENCES↗

Integrated Effects of Site Hydrology and Vegetation on Exchange Fluxes and Nutrient Cycling at a Coastal Terrestrial‐Aquatic Interface

Abstract The complex interactions among soil, vegetation, and site hydrologic conditions driven by precipitation and tidal cycles control the biogeochemical transformations and bi‐directional exchange of carbon and nutrients across the terrestrial–aquatic interfaces (TAIs) in coastal regions. This study uses a highly mechanistic model, Advanced Terrestrial Simulator (ATS)‐PFLOTRAN, to explore how these interactions affect exchanges of materials and carbon and nitrogen cycling. We used a transect in the Chesapeake Bay region that spans zones of open water, coastal wetland, transition, and upland forest. We designed several simulation scenarios to parse the effects of the individual controlling factors and the sensitivity of carbon cycling to reaction rate parameters derived from laboratory experiments. Our simulations reveal an active zone for carbon cycling under the transition zones between the wetland and the upland. Evapotranspiration is found to enhance the exchange fluxes between the surface and subsurface domains, resulting in a higher dissolved oxygen concentration in the TAIs. The transport of organic carbon derived from plant leaves and roots provide an additional source of organic carbon needed for the aerobic respiration and denitrification processes in the TAIs. The variability in reaction rate parameters associated with microbial activities is also found to play a dominant role in controlling the heterogeneity and dynamics of the simulated redox conditions. This modeling‐focused exploratory study enabled us to better understand the complex interactions among soil, water and microbes that govern the hydro‐biogeochemical processes at the TAIs, which is an important step toward representing coastal ecosystems in larger‐scale Earth system models.

54 ENVIRONMENTAL SCIENCES↗

Near-surface modification of defective KTaO 3 by ionizing ion irradiation

The synergistic effect of nuclear (S n ) and electronic (S e ) energy loss observed in some ABO 3 perovskites has attracted considerable attention due to the real possibility to modify various near-surface properties, such as the electronic and optical properties, by patterning ion tracks in the defective near-surface regions. In this study, we show that low-energy ion-induced disordering in conjunction with ionizing ion irradiation (18 MeV Si, 21 MeV Ni and 91.6 MeV Xe) is a promising approach for tailoring ion tracks in the near-surface of defective KTaO 3 . Experimental characterization and computer simulations reveal that the size of these latent ion tracks increases with S e and level of pre-existing damage. These results further reveal that the threshold S e value (S e th ) for track creation increases with decreasing pre-damage level. The values of S e th increase from 5.02 keV nm -1 , for a pre-existing fractional disorder of 0.53 in KTaO 3 , to 10.81 keV nm -1 for pristine KTaO 3 . Above these thresholds, amorphous latent tracks are produced due local melting and rapid quenching. Below a disorder fraction of 0.08 and S e ≤ 6.68 keV nm -1 , the synergistic effect is not active, and damage accumulation is suppressed due to a competing ionization-induced damage annealing process. These results indicate that, depending on S e and the amount of pre-existing damage, highly ionizing ions can either enhance or suppress damage accumulation in KTaO 3 , thus providing a pathway to tailoring defects states. Comprehending the conflicting roles of highly ionizing ions in defective ABO 3 oxides is vital for understanding and predictive modeling of ion-solid interactions in complex oxides, as well as for achieving control over ion track size in the near-surface of defective KTaO 3 .

36 MATERIALS SCIENCE↗

In-situ synthesis of oxides by reactive process atmospheres during L-PBF of stainless steel

Traditionally, reactive gases such as oxygen (O 2 ) and carbon dioxide (CO 2 ) have been avoided during laser powder bed fusion (L-PBF) of metals and alloys based on the notion that it may lead to defect formation and poor properties. Here, in this work, we show that instead, these gases can be used to form sub-μm-sized oxide particles in-situ during the L-PBF process in an Fe-Cr-Al-Ti stainless steel and lead to improved room temperature and high-temperature mechanical properties. We manufactured cube samples using pure Ar and various reactive gas atmospheres, namely an O 2 /Argon (Ar) mixture containing 0.2 % O 2 and CO 2 /Ar mixtures containing up to 100 % CO 2 . Co-axial measurements of infrared radiation emitted from the melt pool showed correlation to the presence of O 2 or CO 2 in the gas mixture. Builds produced under CO 2 -containing atmosphere contained complex oxides with an average diameter of ~40 nm, an Al-rich core and a Ti-rich shell. Due to the high cooling rates typical to L-PBF, agglomeration of oxides and slag formation on the surface of the samples could almost be entirely avoided. Compression tests at temperatures up to 800 °C showed that the samples produced in 100 % CO 2 have about 20 % higher yield stress compared to samples produced in Ar. The paper concludes with a discussion of the formation mechanism of the observed oxides. Our results show that in-situ reactions during additive manufacturing processes are a promising pathway to the synthesis of particle-reinforced alloys.

36 MATERIALS SCIENCE↗

Bayesian learning of orthogonal embeddings for multi-fidelity Gaussian Processes

Uncertainty propagation in complex engineering systems often poses significant computational challenges related to modeling and quantifying probability distributions of model outputs, as those emerge as the result of various sources of uncertainty that are inherent in the system under investigation. Gaussian Processes regression (GPs) is a robust meta-modeling technique that allows for fast model prediction and exploration of response surfaces. Multi-fidelity variations of GPs further leverage information from cheap and low fidelity model simulations in order to improve their predictive performance on the high fidelity model. In order to cope with the high volume of data required to train GPs in high dimensional design spaces, a common practice is to introduce latent design variables that are typically projections of the original input space to a lower dimensional subspace, and therefore substitute the problem of learning the initial high dimensional mapping, with that of training a GP on a low dimensional space. Here in this paper, we present a Bayesian approach to identify optimal transformations that map the input points to low dimensional latent variables. The \projection" mapping consists of an orthonormal matrix that is considered a priori unknown and needs to be inferred jointly with the GP parameters, conditioned on the available training data. The proposed Bayesian inference scheme relies on a two-step iterative algorithm that samples from the marginal posteriors of the GP parameters and the projection matrix respectively, both using Markov Chain Monte Carlo (MCMC) sampling. In order to take into account the orthogonality constraints imposed on the orthonormal projection matrix, a Geodesic Monte Carlo sampling algorithm is employed, that is suitable for exploiting probability measures on manifolds. We extend the proposed framework to multi-fidelity models using GPs including the scenarios of training multiple outputs together. We validate our framework on three synthetic problems with a known lower-dimensional subspace. The benefits of our proposed framework, are illustrated on the computationally challenging aerodynamic optimization of a last-stage blade for an industrial gas turbine, where we study the effect of an 85-dimensional shape parameterization of a three-dimensional airfoil on two output quantities of interest, specifically on the aerodynamic efficiency and the degree of reaction

42 ENGINEERING↗

Improved manufacturability of vacuum thermoforming molds by segmentation

Thin polymer shell components used in consumer goods, the automotive industry, and marine applications are often produced by thermoforming. Depending on the quantity of parts, molding processes can increase throughput significantly when compared to processes such as milling or additive manufacturing. One disadvantage to mold-based manufacturing methods, however, is that molds must be produced to form the required geometry. Mold production is often complex and requires a large capital investment. For large-scale molding processes, mold production is sometimes complicated by the requirement for large working volume machine tools which are capable of machining the cavities to the appropriate tolerances for the selected molding process. Here, this paper focuses on a segmented design for a large draw ratio (>3:1 molded surface area to sheet surface area) thermoforming mold and a cost-benefit analysis of segmentation versus monolithic design. Discussion of the thermoforming tests are included. The primary goals were to improve manufacturability and reduce the overall cost of the mold.

Manufacturability↗

Relating Molecular Properties to the Persistence of Marine Dissolved Organic Matter with Liquid Chromatography–Ultrahigh-Resolution Mass Spectrometry

Marine dissolved organic matter (DOM) contains a complex mixture of small molecules that eludes rapid biological degradation. Spatial and temporal variations in the abundance of DOM reflect the existence of fractions that are removed from the ocean over different time scales, ranging from seconds to millennia. However, it remains unknown whether the intrinsic chemical properties of these organic components relate to their persistence. Here, we elucidate and compare the molecular compositions of distinct DOM fractions with different lability along a water column in the North Atlantic Gyre. Our analysis utilized ultrahigh-resolution Fourier transform ion cyclotron resonance mass spectrometry at 21 T coupled to liquid chromatography and a novel data pipeline developed in CoreMS that generates molecular formula assignments and metrics of isomeric complexity. Clustering analysis binned 14 857 distinct molecular components into groups that correspond to the depth distribution of semilabile, semirefractory, and refractory fractions of DOM. The more labile fractions were concentrated near the ocean surface and contained more aliphatic, hydrophobic, and reduced molecules than the refractory fraction, which occurred uniformly throughout the water column. These findings suggest that processes that selectively remove hydrophobic compounds, such as aggregation and particle sorption, contribute to variable removal rates of marine DOM.

54 ENVIRONMENTAL SCIENCES↗

Responsible Artificial Intelligence for Insider Threat Mitigation

This report examines the application of artificial intelligence (AI) technologies for insider threat mitigation (ITM) programs in nuclear security facilities. Insider threat detection presents unique challenges due to the subtle and adaptive nature of these threats, the complex signatures involved, and the scarcity of available data for analysis. Traditional human-centered approaches, while essential, face limitations in processing large amounts of data continuously and detecting subtle patterns across multiple systems. AI technologies can potentially address these limitations by providing 24/7 monitoring capabilities, identifying complex patterns that might escape human observation, and offering consistent application of security criteria. However, the deployment of AI in nuclear security contexts introduces significant new risks, including workflow disruption, expanded attack surfaces, potential for misuse, and ethical concerns regarding privacy, fairness, transparency, safety, and security. The high-consequence nature of nuclear security decisions demands careful consideration of these risks and systematic approaches to their mitigation.

45 MILITARY TECHNOLOGY, WEAPONRY, AND NATIONAL DEF↗

Additive manufacturing of complex objects using refractory matrix materials

A method for the manufacture of a three-dimensional object using a refractory matrix material is provided. The method includes the additive manufacture of a green body from a powder-based refractory matrix material followed by densification via chemical vapor infiltration (CVI). The refractory matrix material can be a refractory ceramic (e.g., silicon carbide, zirconium carbide, or graphite) or a refractory metal (e.g., molybdenum or tungsten). In one embodiment, the matrix material is deposited according to a binder-jet printing process to produce a green body having a complex geometry. The CVI process increases its density, provides a hermetic seal, and yields an object with mechanical integrity. The residual binder content dissociates and is removed from the green body prior to the start of the CVI process as temperatures increase in the CVI reactor. The CVI process selective deposits a fully dense coating on all internal and external surfaces of the finished object.

Terrani, Kurt A.↗

Liquid Interfacial Electron Microscopy Identifies Nanogalvanic Corrosion in Pearlitic Steel

The nanoscale mechanisms of localized corrosion in low carbon steels have remained elusive due to the complexity of studying the degradative material behavior at nanoscale solid-liquid interfaces. We identified various steps in the nanogalvanic corrosion processes using in-situ liquid-cell scanning transmission electron microscopy (STEM) using a microfluidic holder by Hummingbird Scientific. Initial work, performed at low magnification, identified the initiation point on a 1018 low-carbon steel surface. This initiation point was determined to be a triple junction of two ferrite grains bridging a cementite grain in contact with a baseline electrolyte of 6 uM CO2 dissolved in a buffered (2.78 uM Na2SO4) aqueous solution, pH 6.1. The pre-etched low-carbon steel surface was prepared using focused ion beam lift-out procedures to extract a cross-section of the low-carbon steel surface, which then was thinned to about 150 nm and transferred to a SiN membrane microfluidic window. The transfer was made using a lift-out needle to attach the low-carbon steel lamella to the corner of the SiN window, and then Pt/C deposition held the lamella in contact with the window while it was released from the lift out needle. To identify the triple point on the low carbon steel lamella, prior to attachment on the SiN window, the sample was characterized for compositional variations with energy dispersive x-ray spectroscopy mapping, grain orientation and phase mapping with precession electron diffraction, and thickness mapping with energy filtered transmission electron microscopy. This pre-characterization prior to the in-situ experiment provided a map of the multiphase and multigrain structure, where the in-situ liquid cell imaging provided a clear understanding of the initiation point on the sample. These data were cross-correlated to paint a holistic picture of the triple junction site, enabling low electron-fluence in-situ snapshot imaging to avoid dominating the native corrosion reactions with effects from the incident electron beam. This initial result identified that localized, nanogalvanic corrosion at the phase interface was the dominant corrosion process in the low-carbon steel, so we next targeted the observation of an array of these nanogalvanic features phase boundaries in a pearlite grain. Near-surface ferrite/cementite phase interfaces that typify pearlitic low-carbon steel were extracted, pre-characterized, and imaged for the in-situ corrosion processes. The sample was a cross-section from a pearlite grain, with alternating ferrite and cementite grains that extended microns down from the pre-etched low-carbon steel pipe surface. After contact with a buffered aqueous solution, the phase boundaries between the ferrite and cementite began to dissolve, with observable material loss and thickness changes in the dark-field and bright-field STEM images. Within minutes, the corrosion front proceeded deeper into the material, claiming a thin layer of ferrite around all exposed phase boundaries before progressing laterally into the ferrite matrix, converting the ferrite to corrosion product normal to each buried cementite grain. Formation of the corrosion product causes a volumetric expansion, creating a lateral wedging force that mechanically ejects the cementite grains from their grooves and leaves behind percolation channels into the steel substructure. Rapid and deleterious, this nanogalvanic corrosion pathway represents an important target for understanding and preventing run-away degradation in this common building material. Observation of this corrosion mechanism was enabled by the combination of pre-characterization using standard structural, grain, and compositional analysis in the TEM, which provides maps for understanding the reaction propagation captured in low-dose, in-situ, liquid-cell STEM.

corrosion↗

Additive manufacturing of complex objects using refractory matrix materials

A method for the manufacture of a three-dimensional object using a refractory matrix material is provided. The method includes the additive manufacture of a green body from a powder-based refractory matrix material followed by densification via chemical vapor infiltration (CVI). The refractory matrix material can be a refractory ceramic (e.g., silicon carbide, zirconium carbide, or graphite) or a refractory metal (e.g., molybdenum or tungsten). In one embodiment, the matrix material is deposited according to a binder-jet printing process to produce a green body having a complex geometry. The CVI process increases its density, provides a hermetic seal, and yields an object with mechanical integrity. The residual binder content dissociates and is removed from the green body prior to the start of the CVI process as temperatures increase in the CVI reactor. The CVI process selective deposits a fully dense coating on all internal and external surfaces of the finished object.

Terrani, Kurt A.↗

Peatland hydrology across scale: a probabilistic framework for confronting variability, heterogeneity, and uncertainty (Final Report)

Our project aimed to guide the representations of peatland hydrology and water-carbon feedbacks within Earth System Models, by improving process understanding and developing parsimonious models that elucidate the effects of hydroclimatic variability and spatial heterogeneity. Our research questions centered on the three key drivers of peatland hydrology: (1) seasonal and interannual hydroclimatic fluctuations, (2) spatial heterogeneity of peatland microtopography, and (3) hydrological connectivity across landscape units within a peatland watershed. Our analyses and model testing used existing long-term datasets within the Marcell Experimental Forest (MEF), with new data collected whenever necessary. To date, we have demonstrated the strong influence of water table elevations on the temperature sensitivity of CH 4 emissions, using a newly developed eddy covariance dataset spanning eleven years at MEF. Specifically, higher water tables dampen the increase in CH 4 emissions in the spring as well as their decrease in the fall, resulting in hysteresis. These results imply that any hydroclimatological changes in peatlands that shift seasonal water availability from winter to summer will increase annual CH 4 emissions, even if soil temperature remains unchanged. To further investigate water table elevation change within the peatland microtopography, we installed automated water table gauges across four bog-forest transects (the “lagg”) in two peatland watersheds at MEF, which are hotspots of intense biogeochemical activity. These measurements will give us new information about the extent of lagg expansion and contraction during snowmelt and high intensity rainfall events, as well as the directionality of water and nutrient flow into the bog from the surrounding forests. Finally, we investigated the hydrological connectivity across the peatland watershed complex during early spring – a critical period where water table is elevated from snowmelt – using extended hydrological records from MEF (e.g., snow and frost depths, water table elevations, streamflow). The importance of thermal and wintertime processes within the Energy Exascale Earth System Model (E3SM) was determined through sensitivity analyses. Results from data analyses show that (i) streamflow has decreased over decades likely due to increased evapotranspiration rates (despite no detectable trends in precipitation), (ii) spring streamflow generation is controlled by a sequence of fill-and-spill pathways from the snowpack to the stream through peatland water stores, and (iii) frost depth is a key explanatory variable for the timing and magnitude of streamflow. These results suggest that frost plays an important role in connecting surface water storage in peatlands to stream outlets, and that the hydrological connectivity across the peatland watershed complex can mediate the sensitivity of hydrological responses to climate variations.

54 ENVIRONMENTAL SCIENCES↗

Explainable AI (XAI)-driven vibration sensing scheme for surface quality monitoring in a smart surface grinding process

Local Interpretable and Model-agnostic Explanation (LIME), an explainable artificial intelligence (XAI) approach is adapted to identify the globally important time-frequency bands for predicting average surface roughness (Ra) in a smart grinding process. The smart grinding setup consisted of a Supertech CNC precision surface grinding machine, instrumented with a Dytran piezoelectric accelerometer attached to the tailstock quill along the tangential direction (Y-axis). For every grinding pass, vibration signatures were captured, and the ground truth surface roughness values were recorded using a Mahr Marsurf M300C portable surface roughness profilometer. The roughness values ranged from 0.06 to 0.14 microns over the complete set of experiments. Time-frequency domain spectrogram frames were extracted for each of the vibration signals collected during the grinding process. Convolutional Neural Networks (CNNs) were modeled to predict the surface roughness based on these spectrogram frames and their image augmentations. The best CNN model was able to predict the roughness values with an overall R2-score of 0.95, training R2-score of 0.99, and testing R2-score of 0.81 with only 80 sets of vibration signals corresponding to 4 experiments with 20 trials each. Although the data size is not large enough to guarantee such performance metrics in real-world scenarios, one can extract statistically consistent explanations underlying the relationships these complex deep learning models capture. Further, the LIME methodology was implemented on the developed surface roughness CNN model to identify the important time-frequency bands (i.e., the superpixels of a spectrogram) influencing the predictions. Based on the identified important regions on the spectrogram frames, the corresponding frequency characteristics were determined that influence the surface roughness predictions. The important frequency range based on LIME results was approximately 11.7 to 19.1 kHz. The power of XAI was demonstrated by cutting down the sampling rate from 160 kHz to 30, 20, 10, and 5 kHz based on the important frequency range and considering Nyquist criteria. Separate CNN models were developed for these ranges by only extracting time-frequency contents below their corresponding Nyquist cut-offs. A proper data acquisition strategy is proposed by comparing the model performances to argue the selection of a sufficient sampling rate to capture the grinding process successfully and robustly.

42 ENGINEERING↗