Comparison of several resuspension models against measured data Original Research Article

The graphite dust that will be generated in an HTR/PBMR during normal reactor operation will be deposited inside the primary system and will become radioactive due to sorption of fission products. A significant amount of radioactive dust may be resuspended and released to the environment in case of LOCA. Therefore accurate particle resuspension models are required for HTR/PBMR safety analyses. Thermal–hydraulic safety analyses of HTR/PBMR type reactors are typically performed using computer codes such as FLOWNEX, MELCOR, or SPECTRA. A resuspension model has been implemented in the past into the system code SPECTRA. The purpose of the present paper is twofold: • Firstly, a method of implementation of a resuspension into a system code is presented. • Secondly, two new resuspension models are introduced and the results are compared with the existing Vainshtein and Rock’n Roll resuspension models. In contrast to the existing models which are valid for turbulent flows, the new models are applicable for both laminar and turbulent flow regimes. The following conclusions are drawn from the performed exercise: • The implementation of resuspension model is performed in such a way that it has a general validity for both steady state and transient conditions. • Relatively simple, quasi-static models, such as the NRG3 and NRG4 models are as useful as the more complicated dynamic models for resuspension calculation. Applicability to both laminar and turbulent flow is important for analyses of, for example, the PBMR recuperator, where the flow is largely laminar. • The framework of resuspension modeling built into SPECTRA, due to its flexibility and large amount of user-defined coefficients, may be used to perform a quick check of the newly developed theoretical models. • A key factor in successful resuspension predictions is a good knowledge of the adhesion force and its distribution for dust particles deposited on rough surfaces. Experimental data is needed that will allow to obtain adhesion force distribution for the materials and corresponding surfaces roughness of the components in an actual plant.