Development and validation of an on-line CFD based data assimilation method for different stratification regimes

Bijloos Gunther

Promoter

Meyers Johan, (KULeuven), johan.meyers@mech.kuleuven.be

SCK•CEN Mentor

Camps Johan
johan.camps@sckcen.be
+32 14 33 27 61

Expert group

Crisis Management and Decision Support

PhD started

2016-10-01

Short project description

In order to assess the impact of nuclear and radiological incidents with releases to the atmosphere, dose and dispersion models are used. Such model calculations are indispensable during the licensing phase of nuclear installations, during the emergency planning phase, during the release phase combined with monitoring for on-line impact assessment and response and during the post-release phase for dose reconstruction calculations. At the near range, i.e. typically within the first kilometer around the affected nuclear installation, the atmospheric dispersion is strongly influenced by local effects such as buildings and vegetation. Nuclear installations form often a complex terrain of buildings, chimneys and cooling towers all embedded in a natural environment (along river or coast, forested region, …). This makes dispersion simulations within this range too challenging for classical nuclear and radiological atmospheric dispersion models, such as Gaussian based plume and puff models. However, the near range can be specifically interesting to determine:

  • the maximal impact of potential releases (often close to the release point);
  • the planning of evacuation routes;
  • the potential impact on the operators mitigating an incident;
  • the optimal locations for the installation of monitoring equipment around the site;
  • release rate estimations based on data assimilation.

State-of-the-art research conducted at SCK•CEN in collaboration with the KU Leuven have demonstrated the large potential of computational fluid dynamics (CFD) in simulating the atmospheric dispersion at the near-range. It was demonstrated that a significantly improved accuracy over existing dispersion models can be obtained using Reynolds Averaged Navier-Stokes (RANS) CFD  [1]. Furthermore, simulations using Large Eddy Simulation (LES) turbulence modeling have led to a much better understanding regarding the substantially different time-dependent behavior of the cloud gamma dose rate and the instantaneous concentrations [2]. Although CFD computations are known to be very time-consuming, very recent research showed that the application of model reduction techniques can reduce the computational requirements to a level even suited for emergency response situations. However, a number of challenging research questions remain to be addressed before coming to an operational and well validated model.

[1] Vervecken et al. (2013), Accounting for wind-direction fluctuations in Reynolds-avergaed simulation of near-range atmospheric dispersion. Atmospheric environment, 72, 142-150

[2] Vervecken et al. (2014), Dynamic dose assessment by Large Eddy Simulation of the near-range atmospheric dispersion. Radiological Protection, submitted for publication

Objective

The main goals of the PhD will be twofold:

  1. Research on sound numerical simulation methods for the extension of the present model to non-neutral atmospheric stratification and including vegetation and terrain effects. This also includes the development of specific model reduction techniques for these simulation types to guarantee reasonable calculation times in view of the future development of an operational model.
  2. The validation of the numerical simulations by acquiring a sound and complete experimental dataset around a nuclear facility.

Model development and numerical simulations

The simulation framework that will be used for this PhD-study is a state-of-the-art Pseudo-spectral LES solver SP-WIND, which is currently under development at KU Leuven [3]. This higher-order CFD code is optimized for the fast and accurate evaluation of the wind field in the atmospheric boundary layer under different types of stratification. As such, the code is ideal for the characterization of the wind field and the continuation of the dispersion research at the near-range. However, as a first step, SP-WIND needs to be extended to be able to simulate the transport of radioactive gases and to account for vegetation and terrain effects.  Secondly, continued research on model reduction is essential to account for these new developments. Finally, an optimal estimation methodology for the model parameters needs to be established to simulate the transition between different weather patterns using the reduced model. It goes without saying that a comprehensive literature survey is essential to complete each of these steps successfully.

Validation

Validation of numerical simulations will be a key issue to convince the CFD and nuclear community to further develop and use these type of models for performing assessments in the near range. The collection of a high quality dataset is in this context very important. The release of small but measureable quantities (below any health concern) of radionuclides from nuclear installations such as the BR1 (Ar-41) and the Institute of Radioelements (Xe-133, ..) can be used to create such a dataset. Particularly interesting of these installation is that release quantities are well known and existing fence monitoring around these installations gives continuously data on radionuclide fluency rates and/or dose rates. Additional detectors (radiological and meteorological) could be deployed to complete the dataset. However, thorough calibration (cloud calibration) and data collection issues (representative data and locations, …) should be resolved. The calibration of a "small" detector in combination with a large and irregular 3D source (the cloud) is  challenging.  Monte Carlo simulations, often used as part of detector calibrations for specific geometries, are extremely inefficient in such a case.   Apart from testing the models developed within the context of this Ph.D., this dataset will be unique for the international community to test atmospheric dose and dispersion models at the near range.

[3] Meyers, J. and Meneveau, C., “Optimal turbine spacing in fully developed wind-farm boundary layers,” Wind Energy,Vol. 15, 2012, pp. 305–317.