김성민 교수

Based on discrete element method simulations, we propose a new form of the constitutive equation for granular flows independent of packing fraction. Rescaling the stress ratio μ by a power of dimensionless temperature Θ makes the data from a wide set of flow geometries collapse to a master curve depending only on the inertial number I. The basic power-law structure appears robust to varying particle properties (e.g., surface friction) in both 2D and 3D systems. We show how this rheology fits and

We evaluate the effectiveness of pretrained and fine-tuned large language models (LLMs) for predicting the synthesizability of inorganic compounds and the selection of precursors needed to perform inorganic synthesis. The predictions of fine-tuned LLMs are comparable to─and sometimes better than─recent bespoke machine learning models for these tasks but require only minimal user expertise, cost, and time to develop. Therefore, this strategy can serve both as an effective and strong baseline for

Continuously reinforced concrete pavement (CRCP) performance depends on, among other factors, the characteristics of early developing cracks caused by environmental loads. The primary objective is to evaluate effects of design, materials, and construction variables on the characteristics of cracks in CRCP when subjected to environmental loads. A mechanistic model is developed using finite element formulations. Concrete and longitudinal steel are discretized using the plane strain and the frame e

Continuously reinforced concrete pavement (CRCP) performance depends primarily on early-age cracks that result from changes in temperature and drying shrinkage. Presented is the behavior of the CRCP due to the temperature change obtained by using a three-dimensional finite element model. The nonlinear effects of the bond-slip between concrete and steel and between concrete and base have been studied. Modeling for the curling effect and for the viscoelastic material characteristics also has been

This paper presents the results of a study conducted to explore the use of small-scale models of accelerated pavement testing (APT) devices to evaluate the performance of pavements in conjunction with full-scale tests. The motivation for the study was the availability of a model mobile load simulator (MMLS), which had been built originally to illustrate the operation of the full-scale mobile load simulator (MLS) under design at the time. The scaling requirements will be different depending on wh

The wireless power transmission (WPT) system must guarantee human protection from the electromagnetic (EM) field around the system in proportion to the increase of the transmit power. This paper presents an advanced power control scheme to reduce the EM field. The basic operation of the proposed scheme is the automatic control of the transmit power according to the proximity of the human body within a certain distance from the WPT system. For the initial design, the EM field simulation is carrie

The dynamic stress response of concrete pavements subjected to moving tandem-axle loads of constant amplitude and harmonic and arbitrary variations was investigated. The concrete pavement was modeled using a plate of infinite extent on a viscoelastic foundation. Formulations were developed in the transformed field domain using ( a) a double Fourier transform in space and moving space for moving loads of constant amplitude and for the steady-state response to moving harmonic loads and ( b) a trip

We evaluate the ability of machine learning to predict whether a hypothetical crystal structure can be synthesized and explain those predictions to scientists. Fine-tuned large language models (LLMs) trained on a human-readable text description of the target crystal structure perform comparably to previous bespoke convolutional graph neural network methods, but better prediction quality can be achieved by training a positive-unlabeled learning model on a text-embedding representation of the stru

We determine a constitutive equation for developed three-dimensional granular flows based on a series of discrete element method simulations. In order to capture non-local phenomena, normal stress differences, and secondary flows, we extend a previously proposed granular temperature-sensitive rheological model by considering Rivlin-Ericksen tensors up to second order. Three model parameters are calibrated with the inertial number and a dimensionless granular temperature. We validate our model by