In a nuclear accident, radioactive release source term is the critical factor of nuclear emergency response and accident assessment. The modelling of source inversion based on variational data assimilation (VAR) is capable of balancing the environmental radioactive monitoring data to obtain the global optimal source term. But it could be influenced by the discrepancy between predictions of the atmospheric dispersion model and observations, which is defined as the dispersion model error in this study. In order to reduce this influence, the VAR with the dispersion model error (DME-VAR) is proposed. In the DME-VAR, the dispersion model error is quantified by the error coefficients at every monitoring station. These error coefficients and the release source term are estimated at the same time. For limiting the runtime, the DME-VAR program supports parallel processing. Two sets of wind tunnel experiment data for a typical Chinese nuclear power plant site are used to validate and evaluated the performance of the DME-VAR. The results demonstrate that the DME-VAR effectively estimates the error coefficients, and outperforms the VAR in both release rate estimation and radioactive contamination predicting. Moreover, the runtimes of these verification experiments are all reasonable, even for the application in the nuclear emergency response.
Intelligent and communicating real-time diagnostics for the fight against the Fire Compartments.