Validation of the French-German model for the treatment of atmospheric dispersion in accidental release situations with experimental data.
M. Montfort, O. Isnard, R. Martens, H. Schnadt, Seventh International conference on harmonisation within atmospheric dispersion models for regulatory purposes, 28-31/05/2001, Beligrate, Italy.
Some years ago the French-German Commission for the Safety Problems of Nuclear
Installations (DFK, Deutsch-Französische Kommission) initiated the investigation of the
problems that may occur due to the use of different dispersion models within the
emergency preparedness procedures of different countries in the case of an accident
occurring in a nuclear power plant near a border.
The predicted concentration distributions may differ substantially leading to different
countermeasures to be considered in each country. Actually, the models used for
assessment of atmospheric dispersion in Germany and in France yield different results
although being of the same Gaussian type as different approaches were used for the
distribution parameters σ y and σ z (function of transfer time in France and of distance in
Germany). It was then decided to establish a new model based on general and widely
accepted theory of the atmospheric boundary layer.
With respect to the horizontal parameter σ y his model is based on considerations of
spectra of turbulent energy in the atmosphere and their relation to the standard
deviation of the Gaussian distribution (spectral approach). The approach proposed by
Monin and Yaglom has been retained. It separates the diffusion process into the
diffusion of the puff center of mass of a pollutant (standard deviation of the distribution
of particles denoted σ yc) responsible for movement of the puff as a whole and the
relative diffusion of the pollutant in the puff around its center of mass (standard
deviation denoted σ yb) causing an enlargement of the puff together with a decrease of
the concentration of pollutant.
The approach used for σ yb modeling is the one proposed by Smith and Hay who
assumed an isotropic, homogeneous turbulence and a spatial Gaussian distribution of
the pollutant. The diffusion of the puff centers σ yc is derived from the original Taylor’s
formulation that concerns the diffusion of all particles simultaneously leaving a plane of
infinite dimensions meaning that, at any instant, even the greatest eddies participate in
In the vertical direction, the problem is simpler as the range of existing turbulent eddies
is restricted to the small scale, even if a part of the turbulence does not follow the
similarity theory. With respect to the parameter σ z the similarity approach was then
used. This approach is based on the hypothesis that turbulence and its time scale can
be described as function of the atmospheric boundary layer variables friction velocity
u*, convective velocity scale w*, mixing layer height zi and height above ground z. This
hypothesis together with the results of atmospheric boundary layer experiments and
some parameterizations are used for estimates of σ z. As for sampling times between
10 and 30 minutes it is not necessary to distinguish between the growth of the puff and
the meandering of the puff center, the original Taylor’s formulation is applicable. An
approximation of this formulation has been used applying an empirical law that leads in
accordance to Taylor’s result to an evolution of σ z proportional to the travel time t for
small values of t and to t 1/2 for large travel times:
with standard deviation of the vertical velocity fluctuations,
This work has been done with the collaboration of the GRS and TUV (Germany).