As part of a model evaluation programme, the University of Graz (Austria) recorded data on emissions, immissions and meteorology surrounding the VoestAlpine Stahl GmbH. steel mill in Graz for several intervals of varying length and complexity.

Data on meteorology and emissions were passed on to Schorling & Partner as well as other consulting engineers providing services in the field of atmospheric dispersions. We were given the assignment to feed the measured data into our simulation systems and provide the resulting simulation data, in order to determine the accuracy of the separate models.

To ensure that simulation results were not "tweaked" to better suit the measured immission values and thus convey a false degree of accuracy, no data on immissions were provided to us or any of the other participating groups.

After completion and submission of all requested services, the University of Graz compared the simulation results of all participating groups with the real-life measurement data on immission for the corresponding time intervals.

The following plot shows the immission level (Y) plotted against the distance from the release point (X). The magenta line represents the measured immission, the purple line shows the immission calculated by S&P. The remaining lines show the simulation results from a number of other software systems.

The plot shows clearly that both the shape of the curve and the location and level of the maximum immission value were established very accurately by the S&P model, in this case AIR.LAG, a software model for industrial releases.

Furthermore, only the S&P model does not make the critical error of underestimating immission levels. In a critical environment, an underestimation of the expected immissions can result in late and/or incorrect decisions, with potentially hazardous results for the surrounding population and environment.