High fidelity simulation of hazardous plume concentration time series based on models of turbulent dispersion

摘要

High fidelity simulation of hazardous plume concentration time series is important for military operations analysis and for the first responders to gain insights into the impact of hazardous releases. Concentration realisations are crucial to obtain accurate and realistic estimates of human health effects due to exposure to Chemical, Biological and Radiological (CBR) releases and for testing and evaluation of new sensor models, network topologies, and associated data fusion algorithms. Simulation of concentration fluctuations of a plume dispersing in turbulent atmosphere is a challenging task, usually requiring extensive domain knowledge, advanced mathematical expertise, and sophisticated computing resources. This is due to the fact that stochastic model for these fluctuations cannot be postulated based on any ad-hoc assumptions and should be deduced and aligned with underlying models of turbulent mixing and dispersion. For instance, simply using a simple Gaussian probability density function for concentration time series leads to inconsistency in the underlying concentration field (negative values) and hence is not physically realisable. In a recent publication [1], we described a simplified algorithm to generate concentration time series based on the rigorous framework of turbulent dispersion. In this approach, the key statistical parameters of the distributions are fed from the “ensemble-averaged” dispersion models and our algorithm provides the realisation of associated time series. In the current paper, we improve the previously proposed algorithm and extend it to the case of non steady sources. To test our algorithm, we used the Hazard Prediction and Assessment Capability (HPAC)(Fig. 1), developed by Defence Threat Reduction Agency (DTRA), USA, and an example non-stationary random concentration realisation generated by applying the proposed algorithm to a time-varying mean concentration profile sampled from within the HPAC simulation environme- t is presented (Fig 11).