Rapid prediction for the transient dispersion of leaked airborne pollutant in urban environment based on graph neural networks.

Rapidly predicting airborne pollutant dispersion in urban is vital for ventilation design and evacuation planning. Computational fluid dynamics (CFD) simulations are commonly used to provide accurate predictions, but the computational cost is too high. Although graph neural networks (GNNs) provide fast predictions of flow fields by manipulating unstructured mesh on GPU, they suffer from high memory usage and accuracy decreases when applied to large-scale urban scenes. Moreover, it is difficult for GNNs to learn the coupled relationship between wind field and pollutant concentration field. We propose a multi-objective GNN model as CFD surrogate to rapidly predict the transient dispersion of airborne pollutant under the influence of complex wind field patterns in urban environment. Based on random urban layouts generated by a 2D bin packing algorithm, we employ a validated CFD model to construct a sample dataset of wind fields and concentration fields. We leverage graph pooling and multi-scale feature fusion to improve prediction accuracy, and subgraph partitioning of both wind field and concentration field to reduce GPU memory usage. The results show that our GNN model at its best runs 1-2 orders of magnitude faster than CFD simulation with accuracy evaluation metrics R2=0.92, and achieves 70 % GPU memory reduction.

Susceptibility and exposure risk to airborne aerosols in intra-urban microclimate: Evidence from subway system of mega-cities.

The health of intra-urban population in modern megacities relies largely on the biosafety within the microclimate of subway system, which can be vulnerable to epidemical challenges brought by virus-laden bioaerosols under varying factors. The literature has yet to address the association between the exposure risks to infectious pathogens and the dynamic changes of boundary conditions in this densely populated microclimate. This study aims at characterizing the bioaerosol dispersion, evaluating the exposure risks under various train arrival scenarios and hazard releasing positions in a real-world double-decker subway station. The results provide the evidence for the dominating airflow pattern, bioaerosols dispersion behaviors, exposure risk, and evacuation guidance in a representative microclimate of mega-cities. The tunnel effects of nearby pedestrian passageways are found to be dominating the airflow pattern, leading to the discharging of airborne bioaerosols. At least 60 % increasing of discharging rate of bioaerosol is attributed to the arrival of one or two trains at the subway platform compared with the scenario with no train arriving. Results from risk assessment with improved Wells-Riley model estimate 57.62 % of maximum infectivity probability with no train arriving. Large areas near the source at the platform floor still cannot be considered safe within 20 min. For the other two scenarios where trains arrive at the platform, the maximum probability of infection is below 5 %. Moreover, the majority of train carriages can be regarded as safe zones, as the ventilation across the screen door are mostly directed towards the platform. Additionally, releasing the bioaerosols at the platform floor poses the most severe threats to human health, and the corresponding evacuation strategies are suggested. These findings offer practical guidance for the design of the intra-urban microclimate, reinforcing the need for exposure reduction device or contingency plans, and providing potential evacuation strategy towards improved health outcomes.

Exposure risk and emergency evacuation modeling of toxic gas leakage in urban areas under the influence of multiple meteorological factors.

Frequent toxic gas leakage accidents in urban environments cannot be timely controlled and often cause great harm due to the many factors affecting gas diffusion. In this study, based on the coupling method of the Weather Research and Forecasting Model (WRF) and the open source OpenFOAM software platform, the chlorine gas diffusion process at different temperatures, wind speeds, and wind directions in a chemical laboratory and nearby urban areas in Beijing was studied numerically. A dose-response model was used to calculate chlorine lethality and assess exposure risk at the pedestrian level. To predict the evacuation path, an improved ant colony algorithm, a greedy heuristic search algorithm based on the dose-response model, was applied. The results demonstrated that the combination of WRF and OpenFOAM could consider the effects of various factors such as temperature, wind speed, and wind direction on the diffusion of toxic gases. The direction of chlorine gas diffusion was affected by wind direction, and the range of chlorine gas diffusion was affected by temperature and wind speed. The area of high exposure risk (fatality rate above 40%) at high temperatures was 21.05% larger than that at low temperatures. When the wind direction was opposite the building, the high exposure risk area was 78.95% smaller than that under the building direction. The present work provides a promising approach for exposure risk assessment and evacuation planning for the emergency response to urban toxic gas leakage.

A multi-GPU accelerated virtual-reality interaction simulation framework.

In this paper, we put forward a real-time multiple GPUs (multi-GPU) accelerated virtual-reality interaction simulation framework where the reconstructed objects from camera images interact with virtual deformable objects. Firstly, based on an extended voxel-based visual hull (VbVH) algorithm, we design an image-based 3D reconstruction platform for real objects. Then, an improved hybrid deformation model, which couples the geometry constrained fast lattice shape matching method (FLSM) and total Lagrangian explicit dynamics (TLED) algorithm, is proposed to achieve efficient and stable simulation of the virtual objects' elastic deformations. Finally, one-way virtual-reality interactions including soft tissues' virtual cutting with bleeding effects are successfully simulated. Moreover, with the purpose of significantly improving the computational efficiency of each time step, we propose an entire multi-GPU implementation method of the framework using compute unified device architecture (CUDA). The experiment results demonstrate that our multi-GPU accelerated virtual-reality interaction framework achieves real-time performance under the moderate calculation scale, which is a new effective 3D interaction technique for virtual reality applications.