Research Group of Prof. Dr. M. Griebel
Institute for Numerical Simulation

  author = { P.~Boeker and O.~Wallenfang and F.~Koster and R.~Croce
		  and B.~Diekmann and M.~Griebel and P.~Schulze-Lammers},
  title = { The Modelling of Odour Dispersion with Time-Resolved
  journal = {Agrartechnische Forschung},
  year = {2000},
  volume = {4},
  pages = {E84-E89},
  pdf = { 1},
  note = {refereed electronical journal},
  annote = {refereed article,landwirt},
  abstract = { In connection with the methods used so far, this
		  contribution describes a new approach for the modelling of
		  odour dispersion. Using the aid of example cases, the
		  methodology of odour prognosis with different approaches is
		  explained, and their advantages and disadvantages as well
		  as their limitations are discussed. Particular attention is
		  given to close-range dispersion from odour emission sources
		  with low source heights and a complex fluidic environment.
		  Typical examples of such cases are agricultural sources or
		  biological processing plants (composting, sewage treatment
		  plants). The new dispersion model is a further development
		  of the NaSt3D model with two variants of improved
		  dispersion modelling, an advection-diffusion approach
		  (Euler model) and a Lagrange-particle model. This model is
		  able to conduct time-resolved calculations of flows and
		  dispersion and hence allows the question of concentration
		  fluctuation, which is important for odour phenomena, to be
		  integrated into the model. The parallelizing of the
		  computer code enables terrain- and source configurations
		  which have been too complex thus far to be calculated in a
		  fine division of the calculation grid. At present, computer
		  clusters and high-performance computers can be used for
		  this purpose in anticipation of the fast further
		  development of efficient personal computers. The
		  consistently analytical approach avoids empirical model
		  supplements with adaptation parameters, such as the
		  otherwise necessary models of exceeding probability, and
		  can thus be calibrated on a physical basis.}