National Central University
  • Taoyuan City, Taiwan
Recent publications
Geophysical modelling performs to obtain subsurface structures in agreement with measured data. Freeware algorithms for geoelectrical data inversion have not been widely used in geophysical communities; however, different open-source modelling/inversion algorithms were developed in recent years. In this study, we review the structures and applications of openly Python-based inversion packages, such as pyGIMLi (Python Library for Inversion and Modelling in Geophysics), BERT (Boundless Electrical Resistivity Tomography), ResIPy (Resistivity and Induced Polarization with Python), pyres (Python wrapper for electrical resistivity modelling), and SimPEG (Simulation and Parameter Estimation in Geophysics). In addition, we examine the recovering ability of pyGIMLi, BERT, ResIPy, and SimPEG freeware through inversion of the same synthetic model forward responses. A versatile pyGIMLi freeware is highly suitable for various geophysical data inversion. The SimPEG framework is developed to allow the user to explore, experiment with, and iterate over multiple approaches to the inverse problem. In contrast, BERT, pyres , and ResIPy are exclusively designed for geoelectric data inversion. BERT and pyGIMLi codes can be easily modified for the intended applications. Both pyres and ResIPy use the same mesh designs and inversion algorithms, but pyres uses scripting language, while ResIPy uses a graphical user interface (GUI) that removes the need for text inputs. Our numerical modelling shows that all the tested inversion freeware could be effective for relatively larger targets. pyGIMLi and BERT could also obtain reasonable model resolutions and anomaly accuracies for small-sized subsurface structures. Based on the heterogeneous layered model and experimental target scenario results, the geoelectrical data inversion could be more effective in pyGIMLi, BERT, and SimPEG freeware packages. Moreover, this study can provide insight into implementing suitable inversion freeware for reproducible geophysical research, mainly for geoelectrical modelling.
CdIn 2 S 4 (CIS) has attracted widespread attention due to its structural stability and photoelectric properties, however, it is difficult to recycle when after usage. Carbon nanofibers (CNFs) as a suitable electron acceptor due to its stable physicochemical properties enhanced the mechanical properties and easily to recycle. There are also few reports on applying CIS/CNFs composite as photocatalyst in removing volatile organic compounds (VOCs). In this study, a novel CIS/CNFs composite was synthesized via a simple hydrothermal method. Various characterizations, such as X-ray diffraction, Scanning Electron Microscope, X-ray Photoelectron Spectroscopy and Transmission Electron Microscopy proved the successful synthesis of CIS/CNFs composite and revealed that CNFs grow on the surfaces of CIS connected with three-dimensional (3D) conductive network. Under visible light irradiation, degradation of toluene reached the optimal level of 86% as the CIS doped with 3% CNFs. Furthermore, 95% removal efficiency was achieved as 200 ppm ozone was added into the system and mineralization rate is also improved. The 3D network of CNFs can facilitate the effective separation and transfer of the photogenerated electron-hole pairs, protect CIS core from photo-corrosion and easily be recycled. Ultimately, plausible of ozone-enhanced photocatalytic mechanisms were proposed. Hence, this study presents a new photocatalyst with visible-light driven ozone-enhanced photocatalysis process toward VOCs.
Extensive research on the in-class synchronous learning aspect of the flipped classroom approach (FCA) has rendered favorable results the past decade. However, less attention has been placed on the out-of-class component of the FCA, specifically the drawbacks that may occur, such as low student accountability while watching the pre-assigned video lessons, thus coming to class unprepared, and not using the latest online technological to its full potential. This study addresses these issues by examining the impact of creating a more synchronous/collaborative online out-of-class flipped-class component. Using the latest online technology, a newly proposed flipped group (PFG) was created and then compared to a regular flipped group (RFG), and a traditional class (TC) over 14 weeks. Fifty-four undergraduate business students from a university in Taiwan participated in the study. Mock International English Language Testing System (IELTS) oral pre/post-tests were given to all participants to investigate which teaching approach is more effective. The results showed the PFG significantly improved overall on average, out-preforming the RFG. Surprisingly, the TC significantly performed better than the RFG. In addition, significant differences and correlations also occurred between the PFG and RFG students’ online learning behaviors and objective performances, such as the time spent online viewing the assigned video lessons and the effects on their quiz and final grade scores. This study’s findings support creating a more synchronous/collaborative online learning environment can enhance the out-of-class component of a FCA, and therefore help improve student’s overall oral/aural EFL learning.
This study investigates the impact of assimilating Formosat-7/COSMIC-II (FS7/C2) radio occultation (RO) refractivity data on predicting the heavy rainfall event that occurred in Taiwan on August 13, 2019. This event was characterized by heavy rainfall over the coastal region of central and southwestern Taiwan. Our investigation is performed using the Weather Research and Forecasting-Local Ensemble Transform Kalman Filter. Generally, assimilating the RO data increases the amount of moisture over the northern South China Sea (SCS) and the Pearl River area in southern China. It was expected that assimilating the RO data would improve low-level moisture analysis, given that more RO data are available for the lower atmosphere compared to those from Formosat-3/COSMIC-I. However, our results show that the experiment that does not include the RO data below 3 km facilitates better rainfall prediction over Taiwan in terms of the intensity and location of heavy rainfall. This heavy rainfall event can be attributed to moisture transport from the Pearl River area, where the RO data at the altitude of 3–5 km provide effective moisture enhancement to deepen the high-moisture layer. The experiment using the local spectral width (LSW) to conduct the quality control (QC) also helps improve rainfall prediction. However, such an LSW-based QC procedure tends to reject significant amounts of RO data 3 km above the land. Based on this case study, our results show that the QC procedure brings a larger impact to rainfall prediction than counterparts that adjust the observation error variance. A sophisticated QC procedure should be developed to optimize the impact of low-level RO data.
An earthquake swarm occurred in Haulien, Taiwan, from April 7 to August 31, 2021. The epicenters are in the range from 23°47′ N to 24°04′ N and from 121°25′ E to 121°42′ E. C q ( r ) and C q ( t ) are the generalized correlation integral of r and t , respectively. From the events with local magnitudes ≥ 3 and focal depths ≤ 25 km, C q ( r ) is calculated for the epicentral and hypocentral distribution (using the distance between two events, r ) and C q ( t ) for the time sequence (using the inter-event time between two events, t ). The multifractal dimension D q (q = 2, 3, …, 15) is the slope of the linear portion of the log–log plots of C q ( r ) versus r as well as C q ( t ) versus t . For the epicentral distribution, the linear pattern is in the range 0.5 ≤ log( r ) ≤ 1.3. The measured values of D q are all smaller than 2 that is the spatial dimension and monotonically decreases with increasing q . This indicates that the epicentral distribution of the swarm is multifractal. For the hypocentral distribution, a lack of a wide enough linear pattern on the log–log plot makes the hypocentral distribution be not multifractal. For the time sequence, the log–log plot of C q ( t ) versus t shows a linear pattern in the range 0.5 ≤ log( t ) ≤ 1.0. The values of D q are all smaller than 1 that is the time dimension and monotonically decreases with increasing q , thus suggesting multifractality of the time sequence when t is shorter than the maximum inter-event time.
FORMOSAT-3/COSMIC (F3/C) constellation of six micro-satellites was launched into the circular low-earth orbit at 800 km altitude with a 72-degree inclination angle on 15 April 2006, uniformly monitoring the ionosphere by the GPS (Global Positioning System) Radio Occultation (RO). Each F3/C satellite is equipped with a TIP (Tiny Ionospheric Photometer) observing 135.6 nm emissions and a TBB (Tri-Band Beacon) for conducting ionospheric tomography. More than 2000 RO profiles per day for the first time allows us globally studying three-dimensional ionospheric electron density structures and formation mechanisms of the equatorial ionization anomaly, middle-latitude trough, Weddell/Okhotsk Sea anomaly, etc. In addition, several new findings, such as plasma caves, plasma depletion bays, etc., have been reported. F3/C electron density profiles together with ground-based GPS total electron contents can be used to monitor, nowcast, and forecast ionospheric space weather. The S4 index of GPS signal scintillations recorded by F3/C is useful for ionospheric irregularities monitoring as well as for positioning, navigation, and communication applications. F3/C was officially decommissioned on 1 May 2020 and replaced by FORMOSAT-7/COSMIC-2 (F7/C2). F7/C2 constellation of six small satellites was launched into the circular low-Earth orbit at 550 km altitude with a 24-degree inclination angle on 25 June 2019. F7/C2 carries an advanced TGRS (Tri Gnss (global navigation satellite system) Radio occultation System) instrument, which tracks more than 4000 RO profiles per day. Each F7/C2 satellite also has a RFB (Radio Reference Beacon) on board for ionospheric tomography and an IVM (Ion Velocity Meter) for measuring ion temperature, velocity, and density. F7/C2 TGRS, IVM, and RFB shall continue to expand the F3/C success in the ionospheric space weather forecasting.
Six low Earth orbit (LEO) satellites were launched on June 25th, 2019 for a radio occultation (RO) mission for the FORMOSAT-7/COSMIC-2 (F7/C2) program. The GPS and GLONASS RO signals received by these F7/C2 satellites can be used to retrieve atmospheric and ionospheric parameter profiles for atmospheric and ionospheric research. In order to process the received RO signal, the processing system named Taiwan Radio Occultation Processing System (TROPS) is built. TROPS is developed by National Space Organization, Taiwan Analysis Center for COSMIC, and GPS Science and Application Research Center in Taiwan. The ionospheric products of TROPS are electron density profile, ionospheric scintillation index (S4 index), and absolute total electron content (TEC). S4 index has been calculated on board the satellites and other two products are retrieved by TROPS after the observation data downlink to ground. TEC is the linear integration of electron density along the signal propagation path. The electron density profile is retrieved from the relative TEC when the elevation angle of GNSS satellite is negative from F7/C2 satellite. The absolute TEC is the TEC from GNSS satellite to F7/C2 satellite. The difference between absolute and relative TEC is the TEC with/without differential code bias (DCB) correction. Currently, the data for the electron density profile and absolute TEC are provided by TROPS. Users can obtain the products freely from the internet. In this study, the retrieval method and the preliminary F7/C2 ionospheric TEC products retrieved by TROPS are presented in detail.
An earthquake sequence that occurred in the Hualien area during April 7 to August 30, 2021 has been recognized as two swarms by the Central Weather Bureau. Its earthquakes with M L ≤ 6.2 ( M L = local magnitude) and focal depths ≤ 25 km were located in an area from 23°46’ N to 24°04’ N and from 121°25’ E to 121°42’ E. The Morlet wavelet technique is applied to analyze the dominant periods of temporal variations in numbers of daily events for the earthquake sequence in two magnitude ranges, i.e., M L ≥ 3 and M L ≥ 4. Results show that the dominant periods are 30.8 and 38.0 days when M L ≥ 3; while the dominant period does not exist when M L ≥ 4. The fluctuation analysis technique in the natural time domain is used to study the memory effect in the swarm for two magnitude ranges, i.e., M L ≥ 3 and M L ≥ 4. Calculated results show that the memory effect is stronger for the time sequence of magnitudes than for that of inter-event times and higher for M L ≥ 3 earthquakes than for M L ≥ 4 events. Consequently, only the short-term corrected memory effect was operative in the earthquake sequence of the Hualien swarms.
This paper is concerned with the spreading speed for a nonlocal dispersal vaccination model with general incidence. We first prove the existence and uniform boundedness of solutions for this model by using the Schauder’s fixed point theorem. Then, applying comparison principle, we establish the existence of spreading speed for the infective individuals. According to our result, one can see that the spreading speed coincides with the critical speed of traveling wave solution connecting the disease-free and endemic equilibria. In addition, the diffusion rate of the infected individuals can increase the spread of infectious diseases, while the vaccination rate reduces the spread of infectious diseases.
The opportunities and challenges of omnichannel in retail industry have been widely discussed, yet despite these benefits, the key elements that constitute an effective omnichannel and how customers respond to omnichannel retailing strategies remain unclear. This research conducted online surveys to test the effects of omnichannel elements on various brand experiences and customer retention, considering the moderating role of purchase behavior. The results indicate that omnichannel elements (integration, individualization, and interaction) are generally helpful in retaining customers, through omnichannel elements influence brand experiences differently. In addition, these omnichannel elements have different influences on customer retention due to different purchase behaviors. The findings suggest that retailers can use different omnichannel strategies to attract customers’ purchases and provide insights for practitioners who want to use omnichannel strategies to deliver superior experiences for customers.
Educators have indicated that social approaches to reading such as book talk activities are helpful for promoting students' interest in reading. However, it is not possible for teachers to interact with all students to talk about the books they have read as they have different language proficiency levels and different topics of interest. This study thus aimed to understand the affordances of a chatbot built with artificial intelligence techniques as a book talk companion, and to explore the role of the interaction in students' engagement and interest in reading. Adopting AI techniques, the chatbot in this study had basic understanding of 157 books. While students could choose any of the books to read and interact with the chatbot, the chatbot provided book talk and social affective cues to facilitate the book talk. Multiple data sources from 68 students participating in a 6-week reading activity were collected and analyzed. It was found that students perceived a high level of social connection with the chatbot. In particular, students talking with the chatbot maintained a stable level of situational interest in the value dimension, while the interest of those who did not participate in the book talk with the chatbot faded significantly. Students' perceptions of the social connection with the chatbot were closely related to their engagement in the reading activity and correlated with both their triggered-situational interest and maintained situational interest. The results provide insights into how a chatbot with AI techniques can create a positive reading experience to sustain students’ interest in learning.
This study aimed to develop a technique to chemically characterize odor issues in neighborhoods of designated industrial zones with pronounced emissions of volatile organic compounds (VOCs). Due to the elusive nature of odor plumes, speedy detection with sufficient sensitivity is required to capture the plumes. In this demonstration, proton-transfer-reaction mass spectrometry (PTR-MS) was used as the front-line detection tool in an industrial zone to guide sampling canisters for in-laboratory analysis of 106 VOCs by gas chromatography-mass spectrometry/flame ionization detector (GC-MS/FID). The fast but less accurate PTR-MS coupled with the slow but accurate GC-MS/FID method effectively eliminates the drawbacks of each instrument and fortifies the strength of both when combined. A 10-day PTR-MS field screening period was conducted to determine suitable trigger VOC species with exceedingly high mixing ratios that were likely the culprits of foul odors. Twenty canister samples were then collected, triggered by m/z 43, 61 (ethyl acetate, fragments, EA), m/z 73 (methyl ethyl ketone, MEK), or m/z 88 (morpholine) in all cases. Internal consistency was confirmed by the high correlation of critical species in the PTR-MS and trigger samples. Several long-lived halocarbons were exploited as the intrinsic internal reference for quality assurance. Oxygenated VOCs (OVOCs) accounted for 15%–75% of the total VOC mixing ratios in the triggered samples. However, EA and MEK, the most prominent OVOC species, did not appear to have common sources with morpholine, which presented with PTR-MS peaks incoherent with the other OVOCs. Nevertheless, these distinctive OVOC plumes were consistent with the multiple types of odor reported by the local residents. In contrast with the triggered sampling, random samples in the same industrial zone and roadside samples in a major metropolitan area were collected. The pronounced OVOC content in the triggered samples highlighted the advantage over random grab sampling to address odor issues.
A new innovative approach is essential for early and effective diagnosis of gastric cancer, using promoter hypermethylation of the tumor suppressor, SOCS-1, that is frequently inactivated in human cancers. We have developed an amplification-free fiber optic nanoplasmonic biosensor for detecting DNA methylation of the SOCS-1 human genome. The method is based on the fiber optic nanogold-linked sorbent assay of PCR-free DNA from human gastric tumor tissue and cell lines. We designed a specific DNA probe fabricated on the fiber core surface while the other probe is bioconjugated with gold nanoparticles in free form to allow percentage determination and differentiating the methylated and unmethylated cell lines, further demonstrating the SOCS-1 methylation occurs in cancer patients but not in normal cell lines. The observed detection limit is 0.81 fM for methylated DNA, and the detection time is within 15 min. In addition, our data were significantly correlated to the data obtained from PCR-based pyrosequencing, and yet with superior accuracy. Hence our results provide new insight to the quantitative evaluation of methylation status of the human genome and can act as an alternative to PCR with a great potential.
In recent years, building information modeling and artificial intelligence of things (BIM-AIOTs) in the construction industry have gained much attention. Construction engineers and researchers learn about BIM-AIOT and increase their professional knowledge through internet searches. However, the large amount of information on the internet makes it difficult to find specific information. Although some previous work of BIM-related searches exists, most still search with a combination of keywords or longer terms. This paper utilizes a machine learning model with natural language processing (NLP) technique of bidirectional encoder representations from transformers (BERT) integrated with a mobile chatbot as a question and answer (QnA) system. The dataset used for modeling contained 3334 text paragraphs that shortened to 10,002 questions. The result shows an F1 score of around 65% accuracy, which is acceptable for model prediction. Then, the system verifies to synchronize to the server and user interface. The system works well for information search and offers a supporting automation information system in the construction industry. This study achieved conversational machine understanding and a user-friendly BIM-AIOT integration information searches platform. The proposed system has a reliable research-based information source. It is verified as an effective and efficient way to produce fast decision-making. The system is deemed a future application for research-based problem-solving solutions in Architecture, Engineering, and Construction (AEC).
We used CO2 as a carbon source for conversion to carbon nanofibers through the catalytic hydrogenation reaction on a Ni-Na/Al2O3 catalyst. The Ag⁺ adsorbed on carbon nanofibers could be spontaneously reduced through the oxidation–reduction reaction Ni + 2Ag⁺ →2Ag + Ni²⁺. The adsorbed Ag⁺ was reduced by the electrons released from Ni, forming Ag particles on the carbon surface. The Ag deposited on the magnetic carbon nanofibers was employed to remove contaminants in water, in which 4-nitrophenol was reduced to 4-aminophenol. The apparent rate of 4-nitrophenol reduction was significantly enhanced with increasing Ag concentration in the range of 0.5–2.6 wt%, from 8.5 × 10⁻³ to 1.25 × 10⁻¹ s⁻¹. The negative charge on the surface when Ag particles formed was assumed to be an important factor in enhancing the catalytic rate because of the increases in the coverage and adsorption rate of 4-NP on the catalyst surface.
A novel dry process that prepared in mixing with a pore-forming agent employs a single batch process in production of grinding wheels with the advantage without using resin liquid as an additive binder. The agglomeration of a pore-forming agent can be avoided due to no dressing agent mixing with the superfine diamond abrasives and binders. Bulk mass production can still ensure to fulfill the excellent homogeneity in mixing cross the entire grinding wheel. Therefore, an uneven mixing caused by mixing errors and human factors can be minimized. The grinding wheel can reach an ultrahigh porosity (~82 %) at a lower sintering temperature of 620 °C. Low-temperature sintering can effectively reduce the carbon footprint. A thermogravimetric-differential scanning calorimeter (TG-DSC) was used to determine the thermal properties of superfine frits. Further characterization of ultrafine binders by an optical non-contact dilatometry is utilized to set-up a protocol sintering plan of grinding wheels. Microstructure from scanning electron microscope (SEM) measurement on grinding wheel presents a uniform large pore structure which offers a benefit to wafer surface removal and less scorching during grinding. Atomic force microscopy (AFM), laser scanning digital microscope (LSCM) and SEM results revealed that the ground silicon (Si) wafer had a low surface roughness (~6 nm), without deep scratches, and with a low damaged layer (~0.7 μm) respectively. The wear test on grinding wheels demonstrates it is very cost effective to use a dry processing wheel. This work provides a new method for grinding Si wafers with a new type of wheel being developed by a novel dry superfine diamond grinding technique in a manufacturing process. It is realized that the use of a dry grinding process can save part of chemical polishing process (CMP) and effectively reduces the polishing time.
Background Granular bed filter technology is a newly emerging field, and the science is in its infancy. There is a lack of background information on these consistencies. This study provides new insights into clean processing, which is useful in hot gas cleanup technology. Methods We employed a moving granular bed filter, which is evaluated in two separate performance studies: (1) optimization of particle removal efficiency and bed pressure drop under different movement velocities of silica sand; and (2) high temperature model simulation conducted dust and Silica sand flow rates and superficial velocity. Besides, this evaluated the filter system's dynamic characteristics by measuring variations in the outlet concentration and size distribution of dust particulates. Significant findings The results indicated that the filtered dust in the silica sand considerably affected the dust via inertial impaction and diffusion. With higher movement velocity of silica sand, the higher specific resistance coefficient of k1 and k2 caused better performance of collection mechanism, thereby enhancing the removal efficiency. Consequently, this study observations indicated that at a high movement velocity of silica sand, the gas velocity favors dust attachment on the silica sand particle surface, facilitating dust removal. The purpose of this study is to investigate the efficiency of moving granular bed filter (MGBF) in high-temperature environment under different operation conditions. The effects of the test temperature and movement velocities of silica sand on the dust size distribution and removal efficiency were studied; further, an experimental system comprising a high-temperature environment of 600 °C was established for a filter system. The removal efficiency was enhanced at a test temperature of 25 °C when considering a gas velocity of 500 mm/s and a movement velocity of silica sand of 1.95 mm/s. The test results showed that an improvement in the movement velocity of silica sand decreased the temperature of the bed, increasing the removal efficiency. Furthermore, this study could be used in different high-temperature filter systems for gas cleanup.
Graft copolymers with diblock side-chains Am(-graft-B3Ay)n in a selective solvent have been reported to self-assemble into vesicles, but the structure is expected to differ distinctly from those of lipid bilayers. Surprisingly, the number of alternating hydrophobic A-block and hydrophilic B-block layers in the vesicle can vary from a monolayer to multilayers such as the hepta-layer, subject to the same copolymer concentration. The area density of the copolymer layer is not uniform across the membrane. This structural difference among different layers is attributed to the neighboring environment and the curvature of the layer. Because of the unusual polymer conformations, nonlamellar structures of polymersomes are formed, and they are much more intricate than those of liposomes. In fact, a copolymer can contribute to a single or two hydrophilic layers, and it can provide up to three hydrophobic layers. The influence of the backbone length (m) and side-chain length (y) and the permeation dynamics are also studied. The thickness of hydrophobic layers is found to increase with increasing side-chain length but is not sensitive to the backbone length. Although the permeation time increases with the layer number for planar membranes, the opposite behavior is observed for spherical vesicles owing to the curvature-enhanced permeability associated with Laplace pressure.
Midlatitude stationary waves are relatively persistent large-scale longitudinal variations in atmospheric circulation. Although recent case studies have suggested a close connection between stationary waves and extreme weather events, little is known about the global-scale linkage between stationary waves and wildfire activity, as well as the potential changes in this relationship in a warmer climate. Here, by analysing the Community Earth System Model version 2 large ensemble, we show that a zonal wavenumber 5-6 stationary wave pattern tends to synchronize wildfire occurrences across the Northern Hemisphere midlatitudes. The alternation of upper-troposphere ridges and troughs creates a hemispheric-scale spatial pattern of alternating hot/dry and cold/wet conditions, which increases or decreases wildfire occurrence, respectively. More persistent high-pressure conditions drastically increase wildfire probabilities. Even though the dynamics of these waves change little in response to anthropogenic global warming, the corresponding midlatitude wildfire variability is projected to intensify due to changes in climate background conditions.
Sheet pile wall is often used as a retaining system at the riverbank owing to its economy, convenience, and constructability. The soil deposit nearby river is composed of alluvium soil with a high ground water level. Therefore, the soil deposit usually has high potential of liquefaction. The shaking may induce soil liquefaction when the earthquake occurs, causing the sheet pile wall damage or failure. Each earthquake has different frequency content and acceleration amplitude in real conditions. Thus, it would lead to different behavior of the wall-soil system. In this study, three dynamic centrifuge tests were carried out by NCU geotechnical centrifuge and shaking table under 24 g centrifugal acceleration field. Ottawa sand was used to prepare the liquefiable ground with a prototypical excavation depth of 3 m. The models were subjected to the input motions with different frequency content of 1 Hz and 3 Hz. The horizontal displacement of the sheet pile wall and ground surface were measured by the linear variable differential transformers and surface markers. Test results indicate that the model subjected to input motion with higher 3 Hz content and higher peak base acceleration has higher excess pore water pressure excitation and excitation rate. The shallow layer soil in the backfilled area of it achieved initial liquefaction. Moreover, it also has larger lateral displacement of sheet pile wall and ground surface as compared to the others.KeywordsSheet pile wallDifferent frequency contentLiquefactionDynamic centrifuge test
Institution pages aggregate content on ResearchGate related to an institution. The members listed on this page have self-identified as being affiliated with this institution. Publications listed on this page were identified by our algorithms as relating to this institution. This page was not created or approved by the institution. If you represent an institution and have questions about these pages or wish to report inaccurate content, you can contact us here.
3,883 members
Hsiao-Ting Tseng
  • Department of Information Management
Wen-June Wang
  • Department of Electrical Engineering
Anthony Shiaw-Tseh Chiang
  • Department of Chemical & Materials Engineering
Chia Chu
  • Department of Civil Engineering
Yi-Hung Liu
  • Department of Chemical & Materials Engineering
Taoyuan City, Taiwan