George Xu
Research Subjects
George Xu has a total of 65 co-authors. Here are the top co-authors by number of shared publications:
- Fong Yew Leong
Institute of High Performance Computing Agency for Science Technology and Research Singapore 138632 Singapore
4 shared publications Chang Wei Kang
Institute of High Performance Computing, Agency for Science, Technology and Research, 1 Fusionopolis Way 1 , 138632
3 shared publications- Chin Chun Ooi
Institute of High Performance Computing, Agency for Science, Technology and Research, Singapore, Singapore
3 shared publications - Hee Joo Poh
Fluid Dynamics Department, Institute of High Performance Computing, A*STAR, Singapore
3 shared publications Hongying Li
Institute of High Performance Computing, Agency for Science, Technology and Research, 1 Fusionopolis Way 1 , 138632
3 shared publications- Keng Hui Lim
Institute of High Performance Computing, Agency for Science, Technology and Research, 1 Fusionopolis Way 1 , 138632
3 shared publications
Latest Publications
- 3 November 2026
- Published byย Elsevierย inย Ocean Engineering
- Vol.ย 370
Fixed-weight GlobalโLocal Dual-Memory Operator Inference for ship-motion prediction in time-varying irregular waves
- 3 November 2026
- Published byย Elsevierย inย Ocean Engineering
- Vol.ย 370
Fixed-weight GlobalโLocal Dual-Memory Operator Inference for ship-motion prediction in time-varying irregular waves
- Open Access
- 13 July 2026
- Published byย Elsevierย inย SSRN Electronic Journal
Sea-State-Conditioned Local Expert Operator Inference for Cross-Period Prediction of Ship Motions in Irregular Waves
Existing Operator Inference models accurately predict ship motions within sampled sea states but their accuracy may degrade under unseen wave conditions because a single fixed operator cannot fully represent the variation of system dynamics across different sea states. Building on a previously developed dual-memory non-Markovian Operator Inference framework, this study proposes a sea-state-conditioned local expert Operator Inference method for cross-period prediction of ship motions in irregular waves. The inferred evolution operator is parameterized by peak frequency and significant wave height, enabling the operator to adapt to different sea states while retaining the interpretable dual-memory representation of wave-input history and latent response memory. To reduce the mismatch between training and target conditions, a local expert strategy is introduced by selecting neighboring peak-period groups for operator identification. The framework is evaluated using CFD-generated motion data of a KCS container ship under 24 irregular-wave conditions. A leave-one-peak-period-out strategy is adopted to assess generalization to unseen wave periods. and four model variants are compared. Results show that the proposed local dynamic model achieves the best prediction accuracy, especially in short-period extrapolation cases where fixed-operator models show larger errors.
- 13 July 2026
- Published byย Elsevierย inย SSRN Electronic Journal
Open AccessSea-State-Conditioned Local Expert Operator Inference for Cross-Period Prediction of Ship Motions in Irregular Waves
Existing Operator Inference models accurately predict ship motions within sampled sea states but their accuracy may degrade under unseen wave conditions because a single fixed operator cannot fully represent the variation of system dynamics across different sea states. Building on a previously developed dual-memory non-Markovian Operator Inference framework, this study proposes a sea-state-conditioned local expert Operator Inference method for cross-period prediction of ship motions in irregular waves. The inferred evolution operator is parameterized by peak frequency and significant wave height, enabling the operator to adapt to different sea states while retaining the interpretable dual-memory representation of wave-input history and latent response memory. To reduce the mismatch between training and target conditions, a local expert strategy is introduced by selecting neighboring peak-period groups for operator identification. The framework is evaluated using CFD-generated motion data of a KCS container ship under 24 irregular-wave conditions. A leave-one-peak-period-out strategy is adopted to assess generalization to unseen wave periods. and four model variants are compared. Results show that the proposed local dynamic model achieves the best prediction accuracy, especially in short-period extrapolation cases where fixed-operator models show larger errors.
- Open Access
- 1 September 2023
- Published byย AIP Publishingย inย Physics of Fluids
- Vol.ย 35 ย (9)
Investigating airborne transmission risks: A mathematical model of evaporating droplets with solid residue
The COVID-19 pandemic has spotlit the scientific field of fluid dynamics governing airborne transmission through virus-laden mucosal-salivary droplets. In this work, a mathematical model for airborne droplet dispersion and viral transmission centered on evaporating droplets containing solid residue was proposed. Droplet dynamics are influenced by factors such as initial velocity, relative humidity (RH), and solid residue, in agreement with analytical and experimental results. Interestingly, the maximum droplet dispersion distance depends strongly on initial droplet size and RH, such as 0.8-mm-diameter droplet at 0.3 RH, 1.0 mm at 0.6 RH, and 1.75 mm at 0.9 RH, but only weakly on initial projected velocity. Under realistic conditions, an evaporating sputum droplet can cover a dispersion distance at least three times than that of a pure water droplet. Based on Wells falling curves, the critical droplet size, the largest droplet that can remain suspended in air without settling due to gravity, ranges from 120 ฮผm at 0.3 RH to 75 ฮผm at 0.9 RH. Together, our results highlight the role of evaporation on droplet lifetime, dispersion distance, and transmission risks.
- 1 September 2023
- Published byย AIP Publishingย inย Physics of Fluids
- Vol.ย 35 ย (9)
Open AccessInvestigating airborne transmission risks: A mathematical model of evaporating droplets with solid residue
The COVID-19 pandemic has spotlit the scientific field of fluid dynamics governing airborne transmission through virus-laden mucosal-salivary droplets. In this work, a mathematical model for airborne droplet dispersion and viral transmission centered on evaporating droplets containing solid residue was proposed. Droplet dynamics are influenced by factors such as initial velocity, relative humidity (RH), and solid residue, in agreement with analytical and experimental results. Interestingly, the maximum droplet dispersion distance depends strongly on initial droplet size and RH, such as 0.8-mm-diameter droplet at 0.3 RH, 1.0 mm at 0.6 RH, and 1.75 mm at 0.9 RH, but only weakly on initial projected velocity. Under realistic conditions, an evaporating sputum droplet can cover a dispersion distance at least three times than that of a pure water droplet. Based on Wells falling curves, the critical droplet size, the largest droplet that can remain suspended in air without settling due to gravity, ranges from 120 ฮผm at 0.3 RH to 75 ฮผm at 0.9 RH. Together, our results highlight the role of evaporation on droplet lifetime, dispersion distance, and transmission risks.
- 13 December 2022
- Published byย Elsevierย inย Journal of Wind Engineering and Industrial Aerodynamics
- Vol.ย 232
Automatic selection of release plane for Lagrangian-based wind-driven rain studies
- 13 December 2022
- Published byย Elsevierย inย Journal of Wind Engineering and Industrial Aerodynamics
- Vol.ย 232
Automatic selection of release plane for Lagrangian-based wind-driven rain studies
- Open Access
- 7 September 2022
- Published byย Springer Natureย inย Scientific Data
- Vol.ย 9 ย (1) , 1-7
Geographical clusters of dengue outbreak in Singapore during the Covid-19 nationwide lockdown of 2020
Dengue, a mosquito-transmitted viral disease, has posed a public health challenge to Singaporean residents over the years. In 2020, Singapore experienced an unprecedented dengue outbreak. We collected a dataset of geographical dengue clusters reported by the National Environment Agency (NEA) from 15 February to 9 July in 2020, covering the nationwide lockdown associated with Covid-19 during the period from 7 April to 1 June. NEA regularly updates the dengue clusters during which an infected person may be tagged to one cluster based on the most probable infection location (residential apartment or workplace address), which is further matched to fine-grained spatial units with an average coverage of about 1.35 km2. Such dengue cluster dataset helps not only reveal the dengue transmission patterns, but also reflect the effects of lockdown on dengue spreading dynamics. The resulting data records are released in simple formats for easy access to facilitate studies on dengue epidemics.
- 7 September 2022
- Published byย Springer Natureย inย Scientific Data
- Vol.ย 9 ย (1) , 1-7
Open AccessGeographical clusters of dengue outbreak in Singapore during the Covid-19 nationwide lockdown of 2020
Dengue, a mosquito-transmitted viral disease, has posed a public health challenge to Singaporean residents over the years. In 2020, Singapore experienced an unprecedented dengue outbreak. We collected a dataset of geographical dengue clusters reported by the National Environment Agency (NEA) from 15 February to 9 July in 2020, covering the nationwide lockdown associated with Covid-19 during the period from 7 April to 1 June. NEA regularly updates the dengue clusters during which an infected person may be tagged to one cluster based on the most probable infection location (residential apartment or workplace address), which is further matched to fine-grained spatial units with an average coverage of about 1.35 km2. Such dengue cluster dataset helps not only reveal the dengue transmission patterns, but also reflect the effects of lockdown on dengue spreading dynamics. The resulting data records are released in simple formats for easy access to facilitate studies on dengue epidemics.
- Open Access
- 20 August 2021
- Published byย AIP Publishingย inย Physics of Fluids
- Vol.ย 33 ย (8) , 087118
Risk assessment of airborne COVID-19 exposure in social settings
The COVID-19 pandemic has led to many countries oscillating between various states of lock-down as they seek to balance keeping the economy and essential services running and minimizing the risk of further transmission. Decisions are made about which activities to keep open across a range of social settings and venues guided only by ad hoc heuristics regarding social distancing and personal hygiene. Hence, we propose the dual use of computational fluid dynamic simulations and surrogate aerosol measurements for location-specific assessment of risk of infection across different real-world settings. We propose a 3-tiered risk assessment scheme to facilitate classification of scenarios into risk levels based on simulations and experiments. Threshold values of 840 viral copies and 40% of original aerosol concentration are chosen to stratify low, medium, and high risk. This can help prioritize allowable activities and guide implementation of phased lockdowns or re-opening. Using a public bus in Singapore as a case study, we evaluate the relative risk of infection across scenarios such as different activities and passenger positions and demonstrate the effectiveness of our risk assessment methodology as a simple and easily interpretable framework. For example, this study revealed that the bus's air-conditioning greatly influences dispersion and increases the risk of certain seats and that talking can result in similar relative risk to coughing for passengers around an infected person. Both numerical and experimental approaches show similar relative risk levels with a Spearman's correlation coefficient of 0.74 despite differing observables, demonstrating applicability of this risk assessment methodology to other scenarios.
- 20 August 2021
- Published byย AIP Publishingย inย Physics of Fluids
- Vol.ย 33 ย (8) , 087118
Open AccessRisk assessment of airborne COVID-19 exposure in social settings
The COVID-19 pandemic has led to many countries oscillating between various states of lock-down as they seek to balance keeping the economy and essential services running and minimizing the risk of further transmission. Decisions are made about which activities to keep open across a range of social settings and venues guided only by ad hoc heuristics regarding social distancing and personal hygiene. Hence, we propose the dual use of computational fluid dynamic simulations and surrogate aerosol measurements for location-specific assessment of risk of infection across different real-world settings. We propose a 3-tiered risk assessment scheme to facilitate classification of scenarios into risk levels based on simulations and experiments. Threshold values of 840 viral copies and 40% of original aerosol concentration are chosen to stratify low, medium, and high risk. This can help prioritize allowable activities and guide implementation of phased lockdowns or re-opening. Using a public bus in Singapore as a case study, we evaluate the relative risk of infection across scenarios such as different activities and passenger positions and demonstrate the effectiveness of our risk assessment methodology as a simple and easily interpretable framework. For example, this study revealed that the bus's air-conditioning greatly influences dispersion and increases the risk of certain seats and that talking can result in similar relative risk to coughing for passengers around an infected person. Both numerical and experimental approaches show similar relative risk levels with a Spearman's correlation coefficient of 0.74 despite differing observables, demonstrating applicability of this risk assessment methodology to other scenarios.
- Open Access
- 25 February 2021
- Published byย Springer Natureย inย Scientific Reports
- Vol.ย 11 ย (1) , 1-10
Airborne dispersion of droplets during coughing: a physical model of viral transmission
The Covid-19 pandemic has focused attention on airborne transmission of viruses. Using realistic air flow simulation, we model droplet dispersion from coughing and study the transmission risk related to SARS-CoV-2. Although this model defines most airborne droplets as 8โ16 ยตm in diameter, we infer that larger droplets of 32โ40 ยตm in diameter may potentially be more infectious due to higher viral content. Use of face masks is therefore recommended for both personal and social protection. We found social distancing effective at reducing transmission potential across all droplet sizes. However, the presence of a human body 1 m away modifies the aerodynamics so that downstream droplet dispersion is enhanced, which has implications on safe distancing in queues. At 1 m distance, we found that an average of 0.55 viral copies is inhaled for a cough at median loading, scalable up to 340 copies at peak loading. Droplet evaporation results in significant reduction in droplet counts, but airborne transmission remains possible even under low humidity conditions.
- 25 February 2021
- Published byย Springer Natureย inย Scientific Reports
- Vol.ย 11 ย (1) , 1-10
Open AccessAirborne dispersion of droplets during coughing: a physical model of viral transmission
The Covid-19 pandemic has focused attention on airborne transmission of viruses. Using realistic air flow simulation, we model droplet dispersion from coughing and study the transmission risk related to SARS-CoV-2. Although this model defines most airborne droplets as 8โ16 ยตm in diameter, we infer that larger droplets of 32โ40 ยตm in diameter may potentially be more infectious due to higher viral content. Use of face masks is therefore recommended for both personal and social protection. We found social distancing effective at reducing transmission potential across all droplet sizes. However, the presence of a human body 1 m away modifies the aerodynamics so that downstream droplet dispersion is enhanced, which has implications on safe distancing in queues. At 1 m distance, we found that an average of 0.55 viral copies is inhaled for a cough at median loading, scalable up to 340 copies at peak loading. Droplet evaporation results in significant reduction in droplet counts, but airborne transmission remains possible even under low humidity conditions.
- Top Cited Papers
- 1 November 2020
- Published byย AIP Publishingย inย Physics of Fluids
- Vol.ย 32 ย (11) , 113301
Dispersion of evaporating cough droplets in tropical outdoor environment
The ongoing Covid-19 pandemic has focused our attention on airborne droplet transmission. In this study, we simulate the dispersion of cough droplets in a tropical outdoor environment, accounting for the effects of non-volatile components on droplet evaporation. The effects of relative humidity, wind speed, and social distancing on evaporative droplet transport are investigated. Transmission risks are evaluated based on SARS-CoV-2 viral deposition on a person standing 1 m or 2 m away from the cougher. Our results show that the travel distance for a 100 ยตm droplet can be up to 6.6 m under a wind speed of 2 m/s. This can be further increased under dry conditions. We found that the travel distance of a small droplet is relatively insensitive to relative humidity. For a millimetric droplet, the projected distance can be more than 1 m, even in still air. Significantly greater droplets and viral deposition are found on a body 1 m away from a cougher, compared to 2 m. Despite low inhalation exposure based on a single cough, infection risks may still manifest through successive coughs or higher viral loadings.
- 1 November 2020
- Published byย AIP Publishingย inย Physics of Fluids
- Vol.ย 32 ย (11) , 113301
Top Cited PapersDispersion of evaporating cough droplets in tropical outdoor environment
The ongoing Covid-19 pandemic has focused our attention on airborne droplet transmission. In this study, we simulate the dispersion of cough droplets in a tropical outdoor environment, accounting for the effects of non-volatile components on droplet evaporation. The effects of relative humidity, wind speed, and social distancing on evaporative droplet transport are investigated. Transmission risks are evaluated based on SARS-CoV-2 viral deposition on a person standing 1 m or 2 m away from the cougher. Our results show that the travel distance for a 100 ยตm droplet can be up to 6.6 m under a wind speed of 2 m/s. This can be further increased under dry conditions. We found that the travel distance of a small droplet is relatively insensitive to relative humidity. For a millimetric droplet, the projected distance can be more than 1 m, even in still air. Significantly greater droplets and viral deposition are found on a body 1 m away from a cougher, compared to 2 m. Despite low inhalation exposure based on a single cough, infection risks may still manifest through successive coughs or higher viral loadings.
- 1 August 2020
- Published byย Elsevierย inย Journal of Environmental Chemical Engineering
- Vol.ย 8 ย (4)
Analysis of transport methodologies for pollutant dispersion modelling in urban environments
- 1 August 2020
- Published byย Elsevierย inย Journal of Environmental Chemical Engineering
- Vol.ย 8 ย (4)
Analysis of transport methodologies for pollutant dispersion modelling in urban environments
- 10 May 2019
- Published byย World Scientific Pub Co Pte Ltdย inย International Journal of Computational Methods
- Vol.ย 17 ย (05)
CFD Simulation of Chemical Gas Dispersion Under Atmospheric Boundary Conditions
Pollutant control is one of the key concerns in the design of buildings, for the sake of occupational health, safety and environment sustainability. In particular, risk analyses related to emergency leakage of chemicals from storage tanks or chemical processes have aroused increasing attentions in recent days, as well as the effectiveness of mitigation measures in order to eliminate, reduce and control the risks. In this paper, a CFD methodology with nonreactive chemical gases treated as passive scalars has been developed to simulate the gas dispersion across urban environments, subject to atmospheric boundary layer wind conditions. Special treatments to maintain the consistency in atmospheric boundary layer flow profiles, turbulence modeling and boundary conditions have also been accounted for. The proposed CFD methodology for gas dispersion has been implemented in the open source CFD code โ OpenFOAM. It has been validated by modeling the gas dispersions for two urban-related test cases: the CODASC street canyon test case measured in a laboratory wind tunnel and the Mock Urban Setting Test (MUST) field experiment conducted in the desert area of Utah State. Effects of turbulent Schmidt number (Sct have been primarily addressed in this study. Statistical analyses about the discrepancies between predicted and experimental data have been carried out and statistical performance measures are used to quantify the accuracy of the proposed methodology. Simulations results from passive scalar transport equation demonstrate good agreement with experimental data, though tracer gases heavier than the atmospheric air were used in both the measurements. Furthermore, sensitivity tests also indicate that the accuracy of the simulation results is sensitive to the value of turbulent Schmidt number.
- 10 May 2019
- Published byย World Scientific Pub Co Pte Ltdย inย International Journal of Computational Methods
- Vol.ย 17 ย (05)
CFD Simulation of Chemical Gas Dispersion Under Atmospheric Boundary Conditions
Pollutant control is one of the key concerns in the design of buildings, for the sake of occupational health, safety and environment sustainability. In particular, risk analyses related to emergency leakage of chemicals from storage tanks or chemical processes have aroused increasing attentions in recent days, as well as the effectiveness of mitigation measures in order to eliminate, reduce and control the risks. In this paper, a CFD methodology with nonreactive chemical gases treated as passive scalars has been developed to simulate the gas dispersion across urban environments, subject to atmospheric boundary layer wind conditions. Special treatments to maintain the consistency in atmospheric boundary layer flow profiles, turbulence modeling and boundary conditions have also been accounted for. The proposed CFD methodology for gas dispersion has been implemented in the open source CFD code โ OpenFOAM. It has been validated by modeling the gas dispersions for two urban-related test cases: the CODASC street canyon test case measured in a laboratory wind tunnel and the Mock Urban Setting Test (MUST) field experiment conducted in the desert area of Utah State. Effects of turbulent Schmidt number (Sct have been primarily addressed in this study. Statistical analyses about the discrepancies between predicted and experimental data have been carried out and statistical performance measures are used to quantify the accuracy of the proposed methodology. Simulations results from passive scalar transport equation demonstrate good agreement with experimental data, though tracer gases heavier than the atmospheric air were used in both the measurements. Furthermore, sensitivity tests also indicate that the accuracy of the simulation results is sensitive to the value of turbulent Schmidt number.