Battelle Ingenieurtechnik GmbH Düsseldorfer Straße 9 65760 Eschborn Germany
Telefax: +49 6196 936-199	Telephone: +49 6196 936-0

Company Profile

Battelle Ingenieurtechnik GmbH is an independent private company for contract research and development in Germany. With a staff of approximately 100 employees, Battelle serves industrial and government clients in the development, application and commercialization of new technologies. The key business areas are engineering and R&D services in process technology, energy technology, technical software and information systems. The scope of services comprises software development and application, engineering studies, consultancy, information processing, executing and management of experimental programs in large scale test facilities. Within process engineering, the emphasis is on CFD, structural mechanics, thermal analysis, safety technology and safety analysis.

     Battelle develops CFD programs and program components for the design and optimization of industrial processes and power plants. The software is marketed and installed at industrial end users (petroleum and glass industry, automobile industry, insulation material production) and safety authorities. Likewise, the programs are intensely utilized to provide consultancy and technical advice for industrial clients (chemical industry, energy production, aerospace, safety authorities, etc.).

     In the project BETES, parallel numerical methods are applied to industrial problems and parallel CFD methods are developed in cooperation with the Institut für Schiffbau of the University of Hamburg.

     Expertise exists both in the experimental and numerical investigation of complex flow in complicated geometries (including model development and verification), like flow with combustion and multi-phase flow. To limit computational time to acceptable quantities, parallel combustion and Lagrangian two-phase modules - based on a finite-volume method and domain decomposition - are presently developed. High numerical efficiency is documented both on parallel HPC and on networked computers.

Applications

• Computation of transient flow with combustion
• Coolant liquid flow in a six-cylinder piston engine
• Multiphase flow computation

A combustible gas can unintendedly be set free in buildings, which can be damaged or destroyed if the premixed gases are ignited. Experimental studies in medium and large scale multi-room buildings - filled with a mixture of air and hydrogen - have been carried out to investigate flame propagation and transient pressure loads.

     The experiments revealed that maximum pressure is basically influenced by local gas composition and geometry details. Under adverse conditions, relatively high loads may be encountered even at low gas concentrations.

     To investigate various accident scenarios and the effectiveness of countermeasures, numerical calculations of the transient combustion of premixed gas are employed. As the combustion characteristics are strongly influenced by small scale geometry, computational grids must be sufficiently fine to take into account such details. CPU-time and memory requirements are increased, in particular for three-dimensional applications. Rapid change of flow properties enforce very small time steps, which further increase computational time. Therefore, parallel computations are necessary to enable the calculation of such problems with available computer resources.

	The figure on the left (53kB) gives an impression of the computational mesh for a manufactory hall and two rooms located under the hall. The rooms - connected by doors and vents - are filled with a mixture of combustible gas and air. Boxes and vessels, installed in the rooms, act as obstructions to the flow field.
	After ignition, the flame is dominated by buoyancy, and flame propagation and pressure rise are relatively slow. When the flame surface approaches vents or the wake of obstructions, it enters a region of high turbulence, resulting in vigorous combustion and rapid pressure rise, leading to high loads on the building. (Image: 37kB)

	The design of piston engines demands for a thorough optimization of the thermal management, which is strongly influenced by the coolant liquid flow within the engine. Optimization aims at the best possible temperature distribution in the engine, which is achieved by prescribing the liquid mass flows directed to different regions of the flow field. (Image: 144kB)
	The geometry of the flow field is extremely complicated and irregular (computational mesh and problem specification with courtesy of Mercedes-Benz). Grid generation is grately eased by the employment of unstructured meshes. Local, cell oriented grid refinement assures high numerical resolution in sensible regiois with a minimum of numerical effort. Domain decomposition is the basis of parallel calculations. (Image: 41kB)
	The computed temperature field is shown in the figure on the left (187kB).

The formation of explosive mixtures may occur in pneumatic conveying systems or during production processes, when combustible dust or drops are involved. Typical examples are powdery coal, wood, flour and aluminum as well as finely dispersed drops of oil. Numerical CFD-calculations, based on the Euler/Lagrange method, are used to predict the flow field and the local dust concentration. The results allow to estimate the risk potential and to predict the effectiveness of various countermeasures.

     The Lagrangian method relies on the alternate calculation of the flow field equations for the continuous phase and for the equation of motion for the particulate phase. The coupling between the phases is performed by exchanging coupling terms and repeated application of the two methods until both solutions have converged.

In large devices, the particles must be tracked over long distances, resulting in unadmissibly high computation time. By parallel calculations, they are reduced to acceptable quantities.      The figure (78kB) shows the calculated gas flow field (streamlines and velocity vectors) and particle trajectories in a 12 m³ vessel. Dust is pneumatically fed into the vessel through a central tube. The air - and most of the small particles - leave the vessel through the exit vents. The figure shows a typical calculated trajectory for medium sized particles (d~30 µm).

Another example shows calculated particle trajectories and particle separation in a cyclone (Image: 42kB).