Development and application of a surrogate model for quick estimation of ex-vessel debris bed coolability

•A surrogate model is developed to predict the dryout power of debris bed.•The surrogate model comprises a characteristic factor and Lipinski 0-D model.•Coolability maps are obtained for the prototypical debris bed of Nordic BWR.•The effect of uncertain parameters on non-coolable domain is discussed...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Nuclear engineering and design 2020-12, Vol.370, p.110898, Article 110898
Hauptverfasser: Chen, Yangli, Ma, Weimin
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:•A surrogate model is developed to predict the dryout power of debris bed.•The surrogate model comprises a characteristic factor and Lipinski 0-D model.•Coolability maps are obtained for the prototypical debris bed of Nordic BWR.•The effect of uncertain parameters on non-coolable domain is discussed. During a hypothetical severe accident of a Nordic boiling water reactor (BWR), an ex-vessel particulate debris bed is expected to form in the flooded lower drywell due to melt-coolant interactions after vessel failure. The key parameter to evaluate debris bed coolability is the dryout heat flux (DHF) or dryout power density, representing the limit of heat removal capacity by the coolant. Several numerical codes such as COCOMO have been developed to simulate thermal hydraulics in multi-dimensional debris beds and predict the cooling limit, but they are computationally expensive and not suitable for probabilistic risk analysis. This paper aims to develop a surrogate model which can serve as a quick-estimate tool for the dryout power density of a heap-like debris bed in a saturated water pool. The dryout power density predicted from the COCOMO code is treated as the full model. A characteristic factor is introduced as the dryout power density ratio between the multi-dimensional debris bed (predicted by COCOMO code) and the corresponding one-dimensional debris bed (predicted by Lipinski 0-D model). The characteristic factor is correlated by the Kriging method with six parameters: bed porosity, particle diameter, debris mass, bed slope, cavity radius and containment pressure. After the surrogate model is trained and validated, it is employed to analyze the coolability of prototypical debris beds of a reference Nordic BWR, given the bed mass and containment pressure from MELCOR simulation. Coolability maps are produced as quick look-up diagrams for identification of coolable domain with the variation of porosity, particle diameter and slope angle. A preliminary uncertainty analysis is performed to demonstrate the effect of uncertain input parameters on non-coolable domain.
ISSN:0029-5493
1872-759X
1872-759X
DOI:10.1016/j.nucengdes.2020.110898