Hong Kong University of Science and Technology
Hong Kong University of Science and Technology Institutional RepositoryNot a member yet
162821 research outputs found
Sort by
High-throughput phytoplankton monitoring and screening of harmful and bloom-forming algae in coastal waters with updated functional screening database
No full textClimate change and anthropogenic pressures alter phytoplankton phenology, distribution, and bloom frequency. Healthy phytoplankton communities are crucial for biogeochemical processes, blue carbon sequestration, and climate change mitigation. By employing high-throughput 18S V4 rRNA metabarcoding, we addressed the need for profiling phytoplankton community and response mechanisms in urbanized coastal ecosystems. Through an extensive literature review, we further integrated available databases and updated taxonomic information to construct a functional screening database, comprising 469 algal species identified from literature as toxin-producing or bloom-forming, affecting ecosystem or human health. Results showed an abundance of Mediophyceae and Trebouxiophyceae enriched among the phytoplankton communities in areas with overarching anthropogenic input sources such as estuarine freshwater and wastewater effluents, respectively, with distributions inferred to associate with water current exchanges. The study also expanded current baseline inventory for studied Hong Kong waters, revealing novel harmful algae profiles dominated primarily by Karlodinium veneficum and Cyclotella choctawhatcheeana. We found that harmful algae compositions in various coastal zones were selectively driven by indicators such as salinity, pH, and nitrogen species loading during the wet season. The incorporation of phytoplankton community monitoring and harmful algae screening in this study streamlines and empowers current molecular-based coastal marine surveillance. This not only facilitates baseline monitoring and mechanistic understanding of harmful and bloom-forming algae occurrence patterns but also advances molecular approaches to inform effective management of water resources and mitigation strategies on a global scale.</p
Bio-inspired Multifunctional Soft Robotic Arm for Object Manipulation and Self-Locomotion
No full textInspired by the remarkable grasping and locomotion capabilities of octopus arms, this paper presents the design, fabrication, and experimental evaluation of a bio-inspired soft robotic arm capable of object grasping and bipedal locomotion in both aquatic and terrestrial environments. By integrating soft materials, pneumatic actuation, and control systems, the soft robotic arm replicates key biological features and functions of octopus arms, enabling object grasping and bipedal locomotion in amphibious environments. Experimental results show that the soft robotic arm can grasp objects with diameters ranging from 52 mm to 76 mm, supporting a maximum payload of 185 g in air and 524.5 g underwater. In addition, a pair of robotic arms enables effective underwater bipedal locomotion at 1.51 cm/s. These findings validate the effectiveness of the proposed design and highlight its potential for practical applications in amphibious environments.</p
Synthetic biology and metabolic engineering of microalgae for sustainable lipid and terpenoid production: an updated perspective
No full textMicroalgae are increasingly recognised as powerful platforms for the sustainable production of lipids and terpenoids, with expanding applications in the food, fuel and biomanufacturing industries. In this updated review, we consolidate and critically assess the most recent advances in synthetic biology and metabolic engineering of key microalgal models, including Chlamydomonas reinhardtii, Nannochloropsis spp. and Phaeodactylum tricornutum. We focus on developments that have emerged in the latest waves of research, emphasising novel genetic toolkits that accelerate the Design-Build-Test-Learn (DBTL) cycle, breakthroughs in genome-scale metabolic modelling, and innovative strategies for organelle-targeted biosynthesis of high-value compounds. Recent case studies are compared to highlight trends in successful engineering approaches. By capturing these up-to-date insights, this review outlines the current trajectory of microalgal biotechnology toward scalable, carbon-neutral biofactories for polyunsaturated fatty acids (PUFAs) and diverse terpenoids, reinforcing their role in global sustainability and the circular bioeconomy.</p
Life cycle carbon footprint and sustainability assessment of building-integrated photovoltaics: A comparative review of technologies and applications
No full textAs global efforts to decarbonize the built environment intensify, Building-Integrated Photovoltaics (BIPV) have emerged as a critical technology that combines renewable energy generation with architectural function. This review systematically evaluates the carbon footprint and sustainability performance of BIPV systems using a cradle-to-grave Life Cycle Assessment (LCA) approach. Four major photovoltaic technologies—crystalline silicon, thin-film, organic, and perovskite—are assessed, focusing on environmental trade-offs in energy efficiency, manufacturing emissions, material toxicity, and end-of-life treatment. Case studies of high-performance BIPV projects in cities such as Berlin and Hong Kong illustrate how design integration, contextual adaptability, and technology choice can enhance energy output while reducing operational emissions. The review also highlights notable data gaps—especially for emerging technologies like perovskites—and examines how durability, grid carbon intensity, and recyclability affect life cycle outcomes. By adopting a multi-criteria evaluation framework that balances environmental impact, economic feasibility, and architectural integration, this study provides actionable insights to enhance the scalability and circularity of BIPV systems. The findings emphasize the strategic potential of BIPV to support net-zero energy targets, strengthen urban resilience, and reposition buildings as active participants in the transition to a low-carbon future.</p
Probing selective pollutant removal in nanofiltration processes: Critical insights from separation factor and removal difference
No full textThe nanofiltration (NF) system applied in water and wastewater treatments generally pursues the removal-aimed selectivity, characterized by high removal of pollutants and minimal removal of non-pollutants. The NF system selectivity is traditionally quantified using the separation factor (SF), which represents a fractionation-based selectivity defined as the passage ratio of two solutes through the system. Herein, we demonstrated that SF metric might inadequately evaluate the removal-aimed selectivity in NF systems, and proposed a more relevant metric – the removal difference between pollutants and non-pollutants (ΔR). We performed experimental and modeling analysis to elucidate the impacts of membrane properties (intrinsic water/solute and solute/solute selectivity) and operating conditions (water flux and recovery) on the ΔR. The results show that ΔR offers a promising framework for evaluating the removal-aimed selectivity. Based on this metric, we suggested novel optimization strategies for membrane materials and systems, such as prioritizing membranes with high solute/solute selectivity over those with high water/solute selectivity, alongside rationally reducing water flux and optimizing water recovery. Additionally, multi-pass filtration, as a strategy to enhance solute removal, can only improve ΔR when the increase in pollutant removal is more pronounced than that of non-pollutants. This work provides a deeper understanding of selectivity in pollutant removal, offering a new perspective for the design and optimization of NF-based water treatment processes.</p
Assessing the economic impacts of labour time in autonomous vehicles
No full textPrevious studies have analysed the impacts of the introduction of autonomous vehicles on transport networks and estimated the safety, congestion, freight, parking and vehicle ownership impacts to social welfare and the economy. However, there appears to be a gap in the literature on the economic impacts of individuals allocating travel time in an autonomous vehicle to labour activities. This additional labour time could then have resulting impacts to productivity and the broader economy. This paper addresses this gap through the development of a novel microeconomic model incorporating time use in autonomous vehicles. The model captures an individual’s consumption behaviour, demand for leisure and supply of labour while accounting for the allocation of travel time to labour and leisure. It is an extension of existing microeconomic models of time use for two features: (1) travel utility, and (2) labour while travelling. This model is then implemented in an integrated computable general equilibrium and transport model for Sydney, Australia, and is tested to understand the order of magnitude of impacts. From this model, the increase in autonomous vehicle penetration rate and the resultant increases to household budget from travelling wages will result in a total welfare increase but with a decreasing rate, which are mainly due to congestion effects. Interestingly, the congestion effects also result in the production and value of time first increase, followed by a decrease, which are counter intuitive.</p
Towards high-fidelity urban wind profiles for the built environment: a neural field to fuse multi-source observational data in Guangzhou, China
No full textAccurate urban wind analysis is critically hampered by sparse and heterogeneous observational data. This work presents a solution through NF-MW (stands for Neural Field for Multi-source Winds), a model that fuses data from Doppler LiDAR and wind profiler radar into a continuous high-resolution wind field. By learning a direct mapping from spatio-temporal coordinates to wind values, NF-MW can reconstruct wind speed and direction at any arbitrary height and time. The framework uniquely handles the 360∘ periodicity of wind direction and uses Fourier-enriched features to capture high-frequency gusts and turbulence often missed by other models. In a Guangzhou case study, NF-MW achieved a Mean Absolute Error of 0.55 m/s for wind speed and 8.95∘ for wind direction, demonstrating superior accuracy over traditional methods. This approach provides the building and environment community with a robust method to generate the realistic dynamic wind data essential for applications ranging from pedestrian comfort assessments to urban air quality modeling.</p
Smart predict-then-optimize-based model predictive control for Virtual Power Plants with battery storage
No full textEffective energy management of Virtual Power Plants (VPPs) is key to improving grid flexibility and reducing carbon emissions. VPPs combine photovoltaic systems, battery storage, and flexible loads into a coordinated system. Model Predictive Control (MPC) is widely used to optimize energy dispatch under dynamic pricing and operational constraints. However, existing research predominantly focuses on the impact of prediction model accuracy on MPC performance, often overlooking the importance of aligning forecasting with control objectives. To address this gap, we propose a novel Smart Predict-then-Optimize-based MPC (SPO-based MPC) framework that embeds the VPP dispatch optimization problem involving battery storage into the training of the load forecasting model using differentiable optimization techniques. By aligning the forecasting model with control objectives, the framework enables the generation of predictions that directly support improved control outcomes. The effectiveness of the proposed method is demonstrated through a case study involving two years of real-world data and Time-of-Use (TOU) pricing schemes from three regions: Austin (United States), Cardiff (United Kingdom), and Shenzhen (China). The results indicate that although the SPO-based method may lead to slightly higher prediction errors, it achieves superior control performance. Specifically, the SPO-based MPC reduces total energy costs by approximately 10% on average compared to the conventional MPC approach. This integrated framework enhances the operational efficiency of battery storage and offers a new direction for intelligent energy management systems.</p
Uncertainty-aware digital twins for deep excavations with lateral support
No full textDeep excavations in urban areas involve complex interactions between support structures and highly variable subsurface geomaterials, introducing significant uncertainties that may cause catastrophic failures. Although digital twins (DTs) offer transformative potential for managing uncertainties and controlling risks, existing frameworks lack systematic incorporation of uncertainty, real-time model updating, and spatiotemporal risk assessment through continuous observation assimilation. This paper proposes a DT framework for deep excavations that systematically incorporates uncertainties, particularly those from geomaterials often neglected in DT modeling. The framework delineates geometry, physics, and behaviors of excavation support systems, enabling near real-time model updating by integrating multi-source, time-evolving 3D monitoring data and high-resolution prediction of excavation-induced responses and risks for decision-making. Key components include data-driven ground characterization, uncertainty-aware 3D numerical modeling, physics-guided Bayesian model updating, and quantitative spatiotemporal risk assessment. The framework’s capability for continuous, closed-loop interactions between physical-virtual systems through real-time data exchange is demonstrated using an excavation project.</p
A skillful self-evolving deep-learning framework for pluvial flood process forecasting in urban areas
No full textEffective emergency response to extreme rainstorms requires real-time water depth and velocity information. Yet, this critical need is hampered by conventional hydrodynamic models, which may take from hours to days to compute. While end-to-end surrogate models accelerate flood prediction by directly mapping rainfall processes to flood depth, their accuracy is limited by the highly nonlinear spatiotemporal relationship between the two. To overcome this limitation, we present a self-evolving framework based on two autoregressive convolutional neural networks that iteratively update flood depth and velocity fields at each timestep. The effectiveness of the proposed method in emulating hydrodynamic models is demonstrated through its application to historical storm events (1984–2018) in Hong Kong. The framework simulates a 96-hour flood event in just 1 s over a 79 km2urban area at 30 m resolution, achieving computational speedups of 850 times (GPU-based) and 3,000 times (CPU-based) over the conventional solvers, while maintaining mean absolute errors of 0.0007 m (flow depth) and 0.0003 m/s (flow velocity), which are respectively 16 and 50 % of the errors produced by a comparable end-to-end model. The proposed framework can be applied in any flood-prone area to provide real-time flood early warning and forecasting.</p