城市高架桥对颗粒物扩散分布的影响

近年来,随着中国城市化进程的不断加快,城市高架的数量也与日俱增,但高架桥在缓解交通拥堵的同时也给城市道路空气污染扩散分布带来了新的影响。为了探讨这一新问题,采用实地监测和数值模拟相结合的方法,基于计算流体力学原理(CFD)建立RNG k-ε模型及离散相(DPM)模型,分析对称型街道峡谷内高架桥沿线交通颗粒物的扩散机制。研究结果表明：高架桥的存在会影响街道峡谷内的空气流场,导致桥面颗粒物向桥下地面沉积,造成地面颗粒物浓度增大;高架引桥沿线的交通状态对街谷内湍流动能的分布也有较大影响,进而改变了颗粒物的分布状态,通过模拟发现,车辆湍动能VIT的存在使颗粒物平均浓度相比于未考虑VIT时降低了3.77%,并且加重了高架桥的“盖子效应”,使高架桥面高度以下的颗粒物的平均浓度发生变化,迎风侧浓度增长11.46%~16.52%,而背风侧浓度下降2.82%~8.65%,高架桥源颗粒物在垂直方向上更加容易扩散;同时发现,高架引桥各位置处对应的街谷内颗粒物浓度要高于高架桥面高度处的颗粒物浓度。研究成果可为制定城市高架地区的交通污染控制对策提供理论参考。     

较长周期波浪作用下系泊船舶装卸作业波浪标准的试验研究

针对目前缺乏较长周期波浪作用下系泊船舶装卸作业波浪标准的问题,进行系统的分析研究。采用物理模型试验方法,针对一系列油船、LNG船和散货船船型,考虑较大波浪周期范围,开展水动力特性的模型试验,分析船舶运动量随周期的变化规律,并结合相应规范规定,提出较长周期波浪作用下船舶作业允许波高随周期变化的折减系数。研究成果可为相关研究和设计提供科学支持。     

航道工程三维辅助疏浚设计系统的开发与应用

为提高航道工程疏浚设计的效率和工程量统计精度,在港口工程数字化智能化勘察设计集成系统(HIDAS)部分研究成果的基础上,开发了航道工程三维辅助疏浚设计系统。该系统结合三维地质模型和三维CAD技术,提供参数化智能建模、开挖工程量自动统计、批量生成开挖断面图等功能。将其应用到钦州港东航道扩建工程的疏浚设计过程中,实现了根据不同土质选择不同开挖边坡坡比、不同超深超宽参数自动生成设计断面,分航段、分土类统计开挖工程量、超深工程量、超宽工程量,分航段批量自动生成CAD格式的开挖断面图,大幅度提高了设计效率和质量。     

高铁货物运输安全风险识别与计量

基于快速化货物运输中的安全风险因素,构建了基于马尔科夫链的高铁货运安全可靠度计量分析模型,定量分析基于服务供应链视角的高铁货运各环节的安全可靠度;同时,构建基于贝叶斯网络的高铁货运安全风险评价模型,借助GeNIe软件和数学推理得出根节点的概率重要度,甄别影响高铁货运安全的关键指标与关键环节。最后,通过数值仿真验证了模型的有效性,研究结果表明：高铁货物运输安全可靠度与修复率的相关性较大,加强动车组风险识别和处理能力,可有效提升高铁货物运输安全可靠度;规范高铁货运作业、加强安全防范措施和应急水平是保障高铁货运安全的重要措施。     

人工智能算法在铁道车辆动力学仿真中的应用进展

梳理了人工智能算法在铁道车辆系统动力学仿真中的应用实例和国内外相关文献,概述了铁道车辆动力学仿真中常用的机器学习和深度学习算法,归纳和评述了2种学习算法在铁道车辆系统动力学建模与仿真中的应用分类;从铁道车辆系统动力学建模、动力学性能预测与动力学性能优化等方面入手,详细讨论了人工智能算法应用在力元建模和仿真、轨道不平顺预测、运行平稳性预测、噪声预测、侧风安全性预测、运行安全性预测、悬挂优化、轮轨匹配优化、结构优化以及主动与半主动控制等领域的优势和局限性,指出了现阶段人工智能算法在动力学仿真应用中主要面临的训练样本缺乏、泛化能力不够、可解释性欠缺等问题;展望了今后人工智能算法和车辆系统动力学交叉研究的发展方向和重点研究内容。研究结果表明：融合经典力学和人工智能算法结合的混合建模理论可作为之后的重点研究方向;人工智能算法对解决随机动力学中的随机不确定性,提高随机动力学的性能具有较大的应用潜力;通过人工智能算法与优化算法相结合来实现动力学性能优化,可充分发挥人工智能算法的优势。     

Non-linear modelling and control of semi-active suspensions with variable damping

Electro-hydraulic dampers can provide variable damping force that is modulated by varying the command current; furthermore, they offer advantages such as lower power, rapid response, lower cost, and simple hardware. However, accurate characterisation of non-linear f–v properties in pre-yield and force saturation in post-yield is still required. Meanwhile, traditional linear or quarter vehicle models contain various non-linearities. The development of a multi-body dynamics model is very complex, and therefore, SIMPACK was used with suitable improvements for model development and numerical simulations. A semi-active suspension was built based on a belief–desire–intention (BDI)-agent model framework. Vehicle handling dynamics were analysed, and a co-simulation analysis was conducted in SIMPACK and MATLAB to evaluate the BDI-agent controller. The design effectively improved ride comfort, handling stability, and driving safety. A rapid control prototype was built based on dSPACE to conduct a real vehicle test. The test and simulation results were consistent, which verified the simulation.     

Modelling physical processes of haline stratification in ROFIs with application to the Rhine outflow region

A physical and numerical study is made of the processes governing the stratification and circulation in ROFIs (Regions of Freshwater Influence) where there is an important impact of wind and tides. Observations in the Rhine ROFI showed that the salinity field consists of a mean and a tidally oscillating part. The physical processes are first analysed using the analytical solutions from a one-dimensional two-layer model. A justification is given for the neglection of non-linear advective terms in the equations of momentum and salinity. The dimensionless forms of the solutions can be expressed in terms of a series of dimensionless numbers. It is shown in particular that stratification and cross-shore circulation largely depend on the balance between rotation and turbulent diffusion, which depends in turn on parameters such as the Ekman number, the bottom friction coefficient, the eddy viscosity ratio and the depth of the layer interface. Surface winds either enhance or destroy stratification depending on the wind angle. The response to wind forcing is discussed using classical Ekman theory. To verify the analytical theory numerical tests are performed with a point model including an advanced turbulence closure scheme. Differences arise due to the non-linear interaction between turbulence on the one hand and current shear and stratification on the other hand. It is shown in particular that the amplitude of the tidal forcing and the off-shore horizontal salinity gradient strongly affect the semi-diurnal and semi-monthly variation of stratification. The effect of the wind is found to be in good agreement with the analysis of the two-layer model. Finally, the numerical model is compared with existing observational data in the Rhine ROFI for October 1990.     

Baroclinic eddy formation in a Rhine plume model

A three-dimensional finite difference tidal model, including advective and diffusive transport of salinity, is used in the two-layer model for simulation of Rhine water outflow. Layer depths are adjusted in a way that no advective transports between upper and lower layer take place in case of sufficiently stable stratification.Model results show frontal eddy development related to (limited) growing internal waves in case of weak northeasterly to southeasterly winds. It is shown that baroclinically unstable conditions occur, related to vertical velocity shear, resulting in frontal meanders with wave lengths between 18 and 30 km. Satellite images of sea surface temperature show a comparable behaviour of the temperature front, which is strongly correlated with the salinity front of the Rhine plume.     

Modelling the silica pump in the Permanently Open Ocean Zone of the Southern Ocean

A coupled 1D physical–biogeochemical model has been built to simulate the cycles of silicon and of nitrogen in the Indian sector of the Permanently Open Ocean Zone of the Southern Ocean. Based on a simplified trophic network, that includes two size classes of phytoplankton and of zooplankton, and a microbial loop, it has been calibrated by reference to surface physical, chemical and biological data sets collected at the KERFIX time-series station (50°40′S–68°25′E). The model correctly reproduces the high nutrient low chlorophyll features typical of the studied area. In a region where the spring–summer mixed layer depth is usually deeper than 60 m, the maximum of chlorophyll never exceeds 1.5 mg m⁻³, and the annual primary production is only 68 g C m⁻² year⁻¹. In the surface layer nitrate is never exhausted (range 27–23.5 mmoles m⁻³) while silicic acid shows strong seasonal variations (range 5–20 mmoles m⁻³). On an annual basis 71% of the primary production sustained by nanophytoplankton is grazed by microzooplankton. Compared to North Atlantic, siliceous microphytoplankton is mainly prevented from blooming because of an unfavourable spring–summer light-mixing regime. Silicic acid limitation (high half saturation constant for Si uptake: 8 mmoles m⁻³) also plays a major role on diatom growth. Mesozooplankton grazing pressure excerpts its influence especially in late spring. The model illustrates the efficiency of the silica pump in the Southern Ocean: up to 63% of the biogenic silica that has been synthetized in the photic layer is exported towards the deep ocean, while only 11% of the particulate organic nitrogen escapes recycling in the surface layer.     

Real time assimilation of HF radar currents into a coastal ocean model

A real time assimilation and forecasting system for coastal currents is presented. The purpose of the system is to deliver current analyses and forecasts on based on assimilation of high-frequency radar surface current measurements. The local Vessel Traffic Service monitoring the ship traffic to two oil terminals on the coast of Norway received the analyses and forecasts in real time.A new assimilation method based on optimal interpolation is presented where spatial covariances derived from an ocean model are used instead of simplified mathematical formulations. An array of high frequency radar antennae provides the current measurements. A suite of nested ocean models comprises the model system. The observing system is found to yield good analyses and short range forecasts that are significantly improved compared to a model twin without assimilation. The system is fast, analysis and 6-h forecasts are ready at the Vessel Traffic Service 45 min after acquisition of radar measurements.