DIPLOS (www.diplos.org) is a collaborative project involving several institutions in the UK and France, and aims to improve the modelling of accidental or deliberate releases of air-borne substances in city centres. Recognising that the development and improvement of reliable fast modelling tools require good quality datasets and the development and testing of appropriate parametrizations, DIPLOS addresses these needs by focusing on the following core objectives: (1) To perform detailed wind tunnel experiments and numerical simulations (LES and DNS) of dispersion from continuous and short-duration releases in simulated urban arrays for different geometries and wind directions. (2) To quantify and parametrize the main exchange processes at intersections and in streets of finite extent, including the effects of secondary sources, three-dimensional flow structure, tall buildings and wind direction. (3) To develop empirical and theoretical methods to estimate concentration fluctuations in the short range. (4) To implement the new parametrizations into a network-based dispersion model (SIRANE-RISK), and to evaluate their performance in simulating realistic case studies in central London. This poster will present an overview of the project and first results.

In this research, a large eddy stimulation (LES) model capable of simulating urban areas was developed, and the degree of impact of buildings, parks, and trees on the local temperature distribution was evaluated.

The main features of the LES model include (i)Building resolving, (ii)Roadside trees are resolved in 3-dimensional, (iii) resolving shadows from buildings and trees, (iv)Multiple reflections of short- and long-wave radiation between buildings and trees by radiosity method, and (v) incorporation of cloud physics and atmospheric radiation models (e.g., RRTM).

The radiative environment within an urban canopy layer is an important factor in determining local- or micro-scale temperature distribution. In order to investigate how a 3-D structure (i.e., buildings and trees) can affect the urban thermal environment, we have developed an urban radiation model. Our urban radiation model is able to consider multiple reflections between buildings or trees. Short- and long-wave radiations are calculated by radiosity method. In our tree model, each individual tree is idealized as a porous board constituted by many layers of leaves, and each board is characterized by its Leaf Area Index. The Leaf Area Index is determined by the leaf density of each layer. Optical parameters are leaf transmittance and reflectance. The intensity of direct solar radiation is decreased by passing through the porous boards. Reflected solar radiation is calculated by the radiosity method.

Several model verification tests are performed to evaluate the robustness of model dynamics and physics, and radiation. Based on these numerical test results, our model is correctly developed at least with regarding dynamics, physics, and radiation.

Numerical simulations of thermal environment in Tajimi city, Japan were conducted to perform sensitivity analyses of roadside trees effects, impact evaluations, and future projections of urban thermal environment at city-scale.

Furthermore, we plan to examine how to plan adaptation to urban thermal environmental problems using our LES model.

Harmful gas leakages threaten lives and health of city inhabitants. These gases can release from technological objects such as swimming pools, winter stadiums, cooling chambers or from chemical factories situated in city suburbs, for instance. Hazardous gas leakages usually last less than one hour. Their dispersion differs from the long-duration leakages as their leakage duration belongs to the turbulent part of the atmospheric spectrum. The aim of this paper is to present dispersion from a short-duration ground level point gas source in an idealised urban canopy. It consisted of block of buildings with pitched roofs organised in inner courtyards. Concentration time series were measured by Fast Flame Ionisation Detector in an open low speed wind tunnel specialised in atmospheric boundary layer modelling. Sampler places were placed at the height of human breathing zone. We conducted many experiment realisations under the same mean experimental conditions at each sampler place to obtain statistically representative data sets. From these data sets characteristics of short-duration leakages (arrival time, leaving time, peak concentration, dosage,...) were calculated. We will compare and discuss the characteristics change from replica to replica at individual sampler places within explored model area. Furthermore, we will investigate changes of mean characteristic values of all replicas from place to place on the model.

Wind pressure difference (WPD) between windward and leeward faces is the driving force for natural ventilation of buildings. The magnitude of WPD depends on many parameters, including buildings layout (parallel or staggered), outdoor climate (wind speed and direction, etc.) and building design (orientation, scale, etc.). Focusing on slab layout of residential buildings, this paper presents a fast estimation method for WPD based on Rhino-Grasshopper platform, aiming to evaluate natural ventilation potentials of residential buildings at preliminary design stage. The typical layouts and scales of residential buildings were first analyzed by performing statistic analysis on 469 buildings from 60 communities in Guangzhou and Foshan. The results show that the main range is 0.5~4 for width-to-height, 0.2~2 for depth-to-height, 0.4~1.5 for row-spacing-to-height and 0.2~1 for group-spacing-to height. CFD method was then applied to obtain the WPD of buildings for a series of typical buildings layouts. The relationship between four angles, i.e. horizontal shading angle, vertical shading angle, displacement angle and wind angle, and WPD was analyzed by using multiple regressions, based on which the fast estimation equations for WPD were established. The WPD calculation program was finally developed on Rhino-Grasshopper platform by using the fast estimation equations. The program is believed as an efficient tool to help designers to evaluate natural ventilation potentials of various buildings layouts at preliminary design stage.

Solar radiation heats the wall and surface of canopy, generates a strong buoyancy flow. The impact of this buoyancy is more obvious at the condition of low wind velocity. In a wind tunnel, heating elements were generally used to generate buoyancy. A wind tunnel experiment using a scaled model made of real construction materials was conducted to explore the possibility of applying artificial light as solar radiation. Under the low inflow condition, the wind velocity was easily influenced by the shape of building and buoyancy flow. After heating the roof top, the velocity increased while turbulent intensity decreased. This might make pollutant transporting faster but inhibited pollutants from mixing. We also changed the roof surface properties by applying insulated coating (composed of micro-size hollow silica particles), the velocity decreased while turbulent intensity increased. We assume this might indicate the influence of radiation on wind flow, and thus feasible to introduce radiation in wind tunnel. Using radiation appropriately in wind tunnel can help better representing the different solar angle and shading effects, and understanding the UHI phenomenon. However, there are still many difficulties to overcome, such as measuring the air temperature precisely. Also we should take consideration on the effect of the lamp body, and conduct a better simulation technique for solar radiation study.

When a significant part of the population lives in urban areas with urban heat islands and air pollution becoming real concerns, developing software to simulate the interactions between the atmosphere and buildings is increasingly necessary.

In our research we use the open source CFD software Code_Saturne, developed by EDF R&D (French Electricity Company, Research and Development) with an atmospheric option developed by the CEREA.

Code_Saturne has previously been applied to pollutant dispersion calculations and effects of radiation.

The work presented here will contribute to implementing a building model in Code_Saturne. After introducing our models, we will present the field campaign that we use to compare our simulations to on-site measurements. Then we will show our main results on the influence of building surface temperature on air flows.

Building impacts on the urban atmosphere

Buildings have a dynamic effect on the air flows. They block and divert them, creating localised turbulences. The shape of the building has a huge impact on the flow patterns formed.

They also influence the air flows thermally by heating or cooling the air differentially by conduction and convection.

Code_Saturne and its existing wall models

Code_Saturne is mainly written in Fortran. It uses two simple wall models to model the buildings. The Wall Thermal model represents the wall as a single layer of an equivalent material, whereas in the Force Restore model - first created to model the ground - it is two-layered with a deeper layer at a constant temperature.

BuildSysPro: a building model to couple with Code_Saturne

BuildSysPro is a library written in Modelica language dedicated to building modeling developed by EDF R&D EnerBat.

The purpose is to couple BuildSysPro and Code_Saturne using a matrix modeling of the building dynamics describing the time evolution of the system under the influence of external factors.

The field campaign

EM2PAU was carried out in Nantes (France) in 2011 by the IFSTTAR, the LHEEA, the CSTB and the Acoustic Laboratory of the Université du Maine.

EM2PAU models a canyon made of two blocks of four steel containers, representing a street of aspect ratio 0.7, with various measurements about environmental conditions.

Thermocouples were spread over the installation to get data on wall, ground and air temperatures. An outside anemometer gave the reference air flow, while six sonic anemometers distributed in a vertical plane through the canyon measured local wind speeds. Pyranometers and a pyrgeometer recorded the global and diffuse solar radiation and infrared radiation respectively.

First results

We meshed EM2PAU in order to simulate it with Code_Saturne, and we modeled its containers and the ground for the building model.

We forced the surface temperatures with their measured values for our simulations and we worked with situations presenting different atmospheric characteristics.

First, we studied a situation of neutral atmosphere, when the temperature is the same for ground and air and theoretically has no influence, as observed at approximately 7am and 7pm on April in Nantes. Then we chose moments when the atmosphere was stratified, and therefore influenced by temperature gradients.

For each of these cases, we ran two different simulations. In the first one, temperatures were not taken into account and the CFD calculations considered the atmosphere as neutral with a constant density, whereas the second simulation included thermal effects with the hypothesis of a dry stratified atmosphere.

Getting similar results for these two simulations in the case of neutral atmosphere (at 7am) confirmed our simulation parameters were reasonable before modeling the air flows at different times of the day. Afterwards, we compare these calculation results in situations of stratified atmosphere to highlight the impact of thermal effects.

Dispersion of hazardous substances by events such as big fire or terror in a built-up area causes huge losses of both life and property. For emergency response, decision makers want to know what areas are impacted and how long the impact will last. In order to do this, atmospheric flow information is essential. However, the atmosphere of urban area has a very irregular flow pattern due to surface structures and local thermal imbalance. To deal with this matter, atmospheric flow patterns of large cities are investigated using a Computational Fluid Dynamics(CFD) model. As input data for CFD simulation, fixed log profile assuming neutral condition is created using radiosonde observation data and output from WRF. And CFD model results are compared with surface observation data of 16 points station which installed at Seolleung in Seoul, Korea. Wind analysis using the radiosonde data and WRF model results show similar pattern to the wind observation done at the surface of the urban area.

In CFD analyses of urban area, power law is widely used for inflow boundary condition. Although this law is experimentally obtained, its relevance is assured by previous observations. However, this law is applicable just only for a profile averaged by long enough timescale. For instance, we have found that ten-minute averaged or one-hour averaged profiles which are recorded by a Doppler Lidar System (DLS) still have considerable differences with the power law. Therefore applicability of the power law as inflow boundary condition is limited by the range of averaging timescale. In addition, the air flow at the ground level is still remained to be determined under instantaneous profile (which is far from a smooth curve such as the power law).

We carried out a CFD analysis of an existent urban area in Tokyo. In this analysis we employ actual profiles obtained by DLS as inflow boundary condition. We used instantaneous profiles (ten minutes average) which measured during April to June 2014, and compared two cases;

CASE1: CFD analyses carried out with fluctuated profiles.

CASE2: CFD analysis carried out with totally averaged profiles (in this case, profile is similar to the curve of the power law).

In addition, from result of CASE1, we can know mean and deviations of wind environment due to the change of wind profiles.

Comparing CASE1 and CASE2, for the modeling inflow condition, we can evaluate how the averaging operation affects to the analysis of the air flow at the ground level. From this research, the relevance of the power law is also evaluated in terms of the range of averaging timescale.

Nowadays, CFD analyses are conducted for urban planning or designing. In most cases, actual inflow profiles are hardly available and the power law is still remained as an effective method to model the inflow. However inflow condition affects significantly to computational result of air flow, therefore this research contributes to know the impact of inflow modeling on the urban wind environment.

The population growth in urban centers is the main source of extensive waste generation. Liquid and gaseous waste needing treatment before its disposal as effluent. The wastewater treatment plants are being built ever closer to the sources, in this case, the large urban centers. Specifically, this paper focuses on the discussions about the issue of odorous compounds by sewage treatment plants affecting the urban climate. The odor caused by the presence of these treatment facilities is indicated as a strong source of annoyance to the population. To proceed with proper control of annoyance due to odorous compounds is necessary to know the emission rates of these compounds. The emission calculation methodologies of odorous compounds from quiescent surface (e.g., stabilization ponds) are entirely dependent upon the determination of the compound concentration field on liquid and gaseous phase. For this determination, we use experimental and numerical tools. Experimental measurements is disadvantaged by the thickness of the layer of concentration which must be well known in order to calculate the overall mass transfer coefficient. Thus, numerical methods are important to the progress of these types of studies. The LES methodology was used for the calculation of the velocity field and concentration in the liquid phase (a scale model of a pond belonging to a sewage treatment plant with quiescent surface). The two main dimensionless parameters governing the transport of odorous compounds are Reynolds (Re) and Schmidt (Sc) numbers. The Reynolds number is the ratio between inertial and viscous forces on the fluid. The Schmidt number relates viscosity of the fluid to the diffusivity of the compound in the same fluid. Hydrogen sulphide is the odorant compound studied in this work. This compound has a high Schmidt number. The literature says that for such compounds the transfer is highly dependent on turbulence in the liquid phase. Therefore, in this study, Reynolds number was varied to understand how turbulence affects the calculation of odorous compounds. Reynolds numbers tested were equal to 640, 1280 and 2300, based on friction velocity and it was fixed the value of Schmidt to hydrogen sulphide.

When modeling particle dispersion from discrete sources or in complex geometries, it is often preferable to describe the transport physics from a Lagrangian frame of reference. In turbulent flows where some component of the velocity field is unknown such as in Reynolds-averaged Navier Stokes (RANS) or large-eddy simulation (LES), a model must be employed for the unresolved particle velocity. This is commonly modeled using a modified form of the stochastic Langevin equation. Through a series of papers published in the 1980’s, a general class of Lagrangian stochastic dispersion models were suggested that are physically consistent in that they theoretically adhere to the second law of thermodynamics. However, many workers have reported that in practice the models often fail to satisfy the second law of thermodynamics, and also result in physically unrealistic or ‘rogue’ trajectories. This can be a crippling problem, particularly in complex flows such as urban topologies. We have found that this problem in general arises from cases when the Langevin equation itself becomes unstable regardless of the numerical integration scheme. Physically, this happens when the particle decorrelation timescale turns out to be negative, resulting in an exponentially increasing autocorrelation function. This is an intuitive result if an alternate interpretation of the equation terms is used. When the model is generalized to fully three-dimensional anisotropic turbulence, each particle velocity component becomes coupled, which results in an additional dependence of stability on the chosen discretization. The root causes of this unphysical behavior will be discussed, and several new methods to correct it will be suggested.

In order to investigate the flow, tracer and temperature fields within the obstacle layer in a high resolution, it is often not sufficient to include obstacles by using a parameterisation, but more detailed models are required that account for buildings explicitly. Such prognostic microscale models are commonly applied to assess tracer dispersion influenced by complex building configurations, but also to calculate wind and temperature fields directly. In order to ensure a good quality of the results and - especially in a regulatory context - a common standard, the German Commission of Air Pollution Prevention of VDI/DIN has published an evaluation guideline which tests several properties of these models and compares results for specific test cases with reference data.

The evaluation procedure is separated into several sections. Some are followed by the model developer and some by both, model user and model developer. The general evaluation by the model developer includes requirements for the documentation and realisation of the model. A scientific evaluation covers requirements of the model theory, including the equations, the kind of parameterisation and boundary conditions. For the validation section of the guideline the model developer performs a set of prescribed model runs for the comparison with either other results of the same model, with analytic solutions, or with experimental reference data. The obstacle configurations of the test cases include empty domains without obstacles, different individual obstacles and finally the interaction of several obstacles.

The first version of this guideline (VDI, 2005) has been applied for the evaluation of several microscale models. However, new datasets have since become available and an updated version which takes these advancements into account is currently being prepared. The presented updated guideline is expected to enter a public review process in early 2015.