Most activities of humankind take place in the transition zone between four compartments of the terrestrial system: the unconfined aquifer, including the unsaturated zone; surface water; vegetation; and atmosphere. The mass, momentum, and heat energy fluxes between these compartments drive their mutual state evolution. Improved understanding of the processes that drive these fluxes is important for climate projections, weather prediction, flood forecasting, water and soil resources management, agriculture, and water quality control. The different transport mechanisms and flow rates within the compartments result in complex patterns on different temporal and spatial scales that make predictions of the terrestrial system challenging for scientists and policy makers. The Transregional Collaborative Research Centre 32 (TR32) was formed in 2007 to integrate monitoring with modeling and data assimilation in order to develop a holistic view of the terrestrial system. TR32 is a long-term research program funded by the German national science foundation Deutsche Forschungsgemeinschaft (DFG), in order to focus and integrate research activities of several universities on an emerging scientific topic of high societal relevance. Aiming to bridge the gap between microscale soil pores and catchment-scale atmospheric variables, TR32 unites research groups from the German universities of Aachen, Bonn, and Cologne, and from the environmental and geoscience departments of Forschungszentrum Julich GmbH. Here, we report about recent achievements in monitoring and modeling of the terrestrial system, including the development of new observation techniques for the subsurface, the establishment of cross-scale, multicompartment modeling platforms from the pore to the catchment scale, and their use to investigate the propagation of patterns in the state and structure of the subsurface to the atmospheric boundary layer.

The UNO predicts that in 2030 over 60% of the world population will live in urban environments. This percentage will increase to up to 70% in the year 2050. With that said, more than ever it is essential that we protect the living conditions in inner-urban regions. In this regard the air quality is of particular importance. The microclimatic implications of the densely built-up area mainly influence the dispersal dynamics of pollutants in ambient air. These are in particular fine dust particles and nitrogen oxides [1, 2]. According to the UN as of today only 12% of the world wide urban population are exposed to an air quality that remains under the level set by WHO. According to this survey half of the urban population is exposed to more than two and a half times of the fine dust limit [3]. Major anthropogenic sources of these pollutants are motorized and private transport, the manufacturing industry, power plants and heating systems in private households. Wind field and pollutant dispersion simulations in the urban near-field on the scale of individual buildings or streets have been used for several decades both in science and practice [4]. Here primarily Reynolds Averaged Navier-Stokes (RANS) models are used to give answers to environmental and urban planning questions. However with these models only mean wind fields are represented and the important turbulent exchanges in the near-surface area are fully parameterized. Nowadays the computing power of high-performance computing (HPC) systems allow for more detailed turbulence modelling using Large Eddy Simulation (LES) [5, 6, 7] covering whole city districts with resolutions of up to 1m [8, 9]. On the one hand LES simulations show much better mean flow and concentration distribution and on the other hand provide a reliable prediction of the turbulent fluctuations of the flow characteristics. Vegetation influences are currently considered using simple inventory models covering the leaf area index [10]. In this work we present a LES-LBM based simulation environment using multiple General Purpose GPUs that allows for detailed predictions of the pollutant dispersion in city districts on the micro scale. Also the model respects dynamic traffic loads and local industrial sources of pollution on the building scale. In addition, micro scale partial models describe the influence of the vegetation (e.g. trees) on the pollutant dispersion.

Nuclear magnetic resonance (NMR) relaxometry is a useful tool to estimate transport and storage properties of rocks and soils. However, as there is no unique relation between the NMR signal and these properties in rocks, a variety of empirical models on deriving hydraulic properties from NMR relaxometry data have been published. Complementary to laboratory measurements, this paper introduces a numerical framework to jointly simulate NMR relaxometry experiments and two-phase flow on the micrometer scale. Herein, the NMR diffusion equations were tied to an established Lattice Boltzmann algorithm used in computational fluid dynamics. The numerically simulated NMR data were validated for both surface-limited and diffusion-limited relaxation regimes using analytical solutions available for fully and partially water-saturated simple pore geometries. Subsequently, simulations were compiled using a complex pore space derived from three-dimensional computer tomography (CT) data of an unconsolidated sand and the results were compared to respective NMR T-1 relaxometry data. The NMR transients simulated for different water saturations matched the measured data regarding initial amplitudes (i.e., porosity and saturation) and relaxation behavior (i.e., distribution of water-saturated pores). Thus, we provide a simulation tool that enables study of the influences of structural and physicochemical properties, such as pore connectivity and pore coupling, surface relaxivity, or diffusivity, on partially saturated porous media, e.g, rocks or soils, with NMR T-1 relaxometry data.

Nuclear Magnetic Resonance (NMR) is a useful tool for analyzing gas (e.g., methane) and fluids (e.g., water, oil) in rock formations in order to derive transport and storage parameters such as pore-size distributions or relative permeability. Even though there is considerable NMR data available about hydraulic properties of rock formations, this information is only empirical. To quantify relationships between NMR parameters and transport properties and to assess and improve (numerical) models to deduce transport and storage properties from NMR data on partially saturated rocks we jointly study NMR and multi-phase flow on virtual pore systems derived from micro CT images using an adapted lattice Boltzmann algorithm for multi-phase flow and advection/diffusion processes. To allow a direct joint simulation of the both, geophysical and transport properties of partly saturated soils Lattice Boltzmann (LB) simulations for multi-phase flow and NMR relaxation have been compiled using virtual pore spaces derived from Micro CT images. These numerical experiments show a good quantitative correlation with laboratory observations regarding NMR amplitudes and decay times for varying saturation degrees.

Numerical simulation of building ventilation with a Lattice Boltzmann LES model. In this contribution ventilation effects are studied for the EnergieForum Berlin. To estimate the influence of external wind fields on the ventilation process, wind and temperature fields are simulated with a large eddy simulation (LES) model. With this approach spatially-and time-resolved output is generated. The Lattice Boltzmann method is being used on a D3Q19 grid with a multiple relaxation time (MRT) model. The recently developed EsoTwist scheme, which uses only a single set of LB distributions, is employed. The Smagorinsky model serves as the LES model. To model the temperature dynamics, the HTLBE method is used. The simulations are CUDA-based and carried out on a GeForce GTX580 GPGPU. The decay of the tracer-gas concentration over time and the mean flow through the building generated by temperature gradients and external wind is determined.

In this work, the suitability of the lattice Boltzmann method is evaluated for the simulation of subcritical turbulent flows around a sphere. Special measures are taken to reduce the computational cost without sacrificing the accuracy of the method. A large eddy simulation turbulence model is employed to allow efficient simulation of resolved flow structures on non-uniform computational meshes. In the vicinity of solid walls, where the flow is governed by the presence of a thin boundary layer, local grid-refinement is employed in order to capture the fine structures of the flow. In the test case considered, reference values for the drag force in the Reynolds number range from 2000 to 10 000 and for the surface pressure distribution and the angle of separation at a Reynolds number of 10 000 could be quantitatively reproduced. A parallel efficiency of 80% was obtained on an Opteron cluster.

Transitional flows are difficult to address by Reynolds Averaged Navier-Stokes (RANS) simulations as the spectrum is typically not fully developed. In this work the suitability of the lattice Boltzmann method is evaluated for the simulation of transitional flows. Special measures are taken to reduce the computational cost without sacrificing the accuracy of the method. A large eddy simulation turbulence model is employed to allow efficient simulation of the resolved flow structures on relatively coarse computational meshes. In the vicinity of solid walls, where the flow is governed by the presence of a thin boundary layer, local grid-refinement is employed in order to capture the fine structures of the flow. The lattice Boltzmann code is run on an Opteron cluster. In the considered test case, the pressure distribution and the drag force on a sphere are computed in the Reynolds number range 1000 to 10000 and a parallel efficiency of 80% is obtained.

In this paper we present a new efficient approach for radiative heat transfer simulations for various applications in engineering, combining existing approaches from different fields of computer science and heat transfer. For these application fields we assume radiative exchange between gray, diffuse surfaces in a radiatively nonparticipating medium. To solve the complex space-time behavior of radiation in 3D domains we use the hierarchical radiosity method. Here, the basic idea is to hierarchically subdivide surfaces forming a quad-tree structure until a refinement criterion is reached. The fundamental underlying operation of the radiosity method is visibility detection which can be solved efficiently by using a space partitioning approach for the input surfaces. For this reason we choose a kd-tree which is the best known acceleration structure for visibility detection on irregularly distributed surfaces. These approaches dramatically decrease the complexity of the radiation problem from n(3) to O((k(2) + n) logk), where k is the number of input surfaces and n is the number of refined surfaces. We validate the approach for several non-trivial examples and demonstrate that the scheme is second-order accurate. Copyright (C) 2011 John Wiley & Sons, Ltd.

Two multi-thread based parallel implementations of the lattice Boltzmann method for non-uniform grids on different hardware platforms are compared in this paper: a multi-core CPU implementation and an implementation on General Purpose Graphics Processing Units (GPGPU). Both codes employ second order accurate compact interpolation at the interfaces, coupling grids of different resolutions. Since the compact interpolation technique is both simple and accurate, it produces almost no computational overhead as compared to the lattice Boltzmann method for uniform grids in terms of node updates per second. To the best of our knowledge, the current paper presents the first study on multi-core parallelization of the lattice Boltzmann method with inhomogeneous grid spacing and nested time stepping for both CPUs and GPUs.

In this article we present a software prototype for transient simulations of heat transport problems in civil engineering, combining efficient numerical methods with a computational steering framework which supports interactive modeling, simulation and visualization of heat transfer in complex buildings. For the full radiant energy exchange in threedimensional domains an optimized hierarchical radiosity method is used. Here the radiation is directly coupled to the heat conduction calculation based on the Finite-Difference-Method.

Since the NVIDIA Compute Unified Device Architecture (CUDA) SDK has been released, multiple publications have demonstrated that simulations in the field of Computational Fluid Dynamics (CFD) may substantially benefit from the utilization of General Purpose Graphics Processing Units (GPGPU), which offer a high performance vs. price ratio compared to classical CPUs typically used in PC clusters. The lattice Boltzmann equation for fluid dynamics is derived from the continuous Boltzmann equation, which describes the probability of finding a fluid particle in a certain velocity state at a certain position and at a certain time. The classical Lattice Boltzmann Method (LBM) is structurally restricted to run on uniform Cartesian grids with a single global time step only. It is obviously desirable to change the resolution of the method from fine grids and short time steps in regions of high interest to coarse grids and long time steps in regions of low interest, typically the peripheral regions of the flow field. Grid refinement approaches mainly have been developed for CPU implementations and do not automatically lead to good performance on GPGPUs. A local grid refinement with low data transfer, a compact memory access pattern and little exceptions at the edges and vertices of the interface is desirable. We present a three dimensional multi scale lattice Boltzmann implementation based on a uniform GPGPU code [1]. For the grid refinement, we use the method of [2] which basically preserves the locality of the LBM and does not lead to a performance degradation on GPGPUs. Our study will thus also address the performance of the multi scale implementation in relation to the uniform grid code.

In this paper we describe an extension of a recently developed lattice Boltzmann method for solving the advection–diffusion equation. Our proposed approach allows to couple grids of different grid resolutions and includes a staggered timestepping scheme, interpolations in space and time and finally a scaling step ensuring the continuity of the desired macroscopic quantities across the grid interface.

The fact that the classic lattice Boltzmann method is restricted to Cartesian Grids has inspired several researchers to apply Finite Volume [Nannelli F, Succi S. The lattice Boltzmann equation on irregular lattices. J Stat Phys 1992;68:401-7; Peng G, Xi H, Duncan C, Chou SH. Finite volume scheme for the lattice Boltzmann method on unstructured meshes. Phys Rev E 1999;59:4675-92; Chen H. Volumetric formulation of the lattice Boltzmann method for fluid dynamics: basic concept. Phys Rev E 1998;58:3955-63] or Finite Element [Lee T, Lin CL. A characteristic Galerkin method for discrete Boltzmann equation. J Comp Phys 2001;171:336-56; Shi X, Lin J, Yu Z. Discontinuous Galerkin spectral element lattice Boltzmann method on triangular element. Int J Numer Methods Fluids 2003;42:1249-61] methods to the Discrete Boltzmann equation. The finite volume method proposed by Peng et al. works on unstructured grids, thus allowing an increased geometrical flexibility. However, the method suffers from substantial numerical instability compared to the standard LBE models. The computational efficiency of the scheme is not competitive with standard methods.We propose an alternative way of discretizing the convection operator using an upwind scheme, as opposed to the central scheme described by Peng et al. We apply our method to some test problems in two spatial dimensions to demonstrate the improved stability of the new scheme and the significant improvement in computational efficiency. Comparisons with a lattice Boltzmann solver working on a hierarchical grid were done and we found that currently finite volume methods for the discrete Boltzmann equation are not yet competitive as stand alone fluid solvers. (C) 2005 Elsevier Ltd. All rights reserved.

Scouring is a phenomenon that is critical to the stability of constructions in the field of hydraulic engineering. While usually empirical formulas are employed for scour prediction, quantitative predictions based on numerical simulations are still not widely successful (1). In the talk an approach to simulate erosion, settling and sediment transport phenomena is presented. The fluid is simulated using a lattice Boltzmann method. To obtain sufficient resolution of small scale structures of obstacles or the river bed with high computational efficiency, hierarchically refined LB grids are used. A sediment model has to handle the transport of sediment in the fluid domain, which is usually (1) modelled as an advection- diffusion equation where the parameters are obtained from the fluid simulation, as well as the morphological change of the river bed, which is the boundary of the fluid domain, due to settling and erosion processes. By using a lattice Boltzmann type scheme for the advection-diffusion equation, we obtain good conservation properties and, as the com- putational grid for the sediment coincides with the one for the fluid, we avoid coupling issues. However, the problem of transition between grids of different resolution has to be addressed. Settling and erosion are handled with phenomenological formulas and a surface tracking algorithm. References