عدم قطعیت و تجزیه و تحلیل حساسیت از آزمون سفر پمپ Kozloduy با استفاده از سینتیک کد 3D هیدروترمال توأم

The modeling of complex transients in nuclear power plants (NPP) remains a challenging topic for best estimate three-dimensional coupled code computational tools. This technique is, nowadays, extensively used since it allows decreasing conservatism in the calculation models and performs more realistic simulation and more precise consideration of multidimensional effects under complex transients in NPPs. Therefore, large international activities are in progress aiming to assess the capabilities of coupled codes and the new frontiers for the nuclear technology that could be opened by this technique. In the current paper, a contribution to the assessment and validation of coupled code technique through the Kozloduy VVER100 pump trip test is performed. For this purpose, the coupled RELAP5/3.3-PARCS/2.6 code is used. The code results were assessed against experimental data. Deviations between code predictions and measurements are mainly due to the used models for evaluating and modeling of the Doppler feedback effect. Further investigations through the use of two “antagonist” uncertainty GRS and the CIAU methods, were considered in order to evaluate and quantify the origin of the observed discrepancies. It was revealed on one hand that relative error quantification discrepancies exist between the two approaches, and further enhancements for both methods are needed.

The evaluation of complex phenomena in NPPs is closely related to the ability of determining the time-space core flux distribution as well as the flow field conditions and the associated effects from heat sources and heat sinks throughout the reactor coolant system. The recent availability of powerful computer and computational techniques has enlarged the capabilities of getting better realistic simulations of complex phenomena including multidimensional effects in NPPs. The application of coupled computer codes was recently identified as an area of high collective interest, especially for nuclear plant design and safety (D’Auria et al., 2004). Therefore, large international activities are in progress aiming to assess the capabilities of coupled codes and the new frontiers for the nuclear technology that could be opened by this technique. This later could be applied for different purposes. A typical example is the coupling of primary system thermal–hydraulic codes with 3D neutron kinetics codes. Other cases include coupling of primary system thermal–hydraulics with structural mechanics, computational fluid dynamics (CFD), nuclear fuel and containment behaviors. The capabilities of the coupled code calculations in simulating, in a best estimate (BE) way, nuclear plant behavior under a wide variety of transient and accident conditions have been largely investigated through several international programs. To pursue this goal, a series of code qualification processes is carried out. This could be performed, for instance, through the consideration of experimental data issued from operational NPP data. In the current framework, a validation of the coupled code technique against a well-documented Kozloduy VVER1000 pump trip is investigated (Vanttola et al., 2005). The transient under consideration concern a relatively common operational transient and could be well-simulated using a best estimate thermal–hydraulic system code. However, the huge amount of experimental data gathered during the test including 2D radial power distribution makes it valuable for coupled code validation purposes. Results of calculation were assessed against experimental data and also through the code-to-code comparisons. However, computer codes, as any simulation tool, are affected by errors arising from the unavoidable approximations connected with the modeling requirements and limitations (Bousbia Salah, 2004). Thus, sensitivity and uncertainty analyses must be carried out to supplement the code results. In order to identify and assess the observed deviations between the measurement and the coupled code calculations a series of uncertainty and sensitivity calculations using statistical GRS (Langenbuch et al., 2005) and the deterministic CIAU (D’Auria and Giannotti, 2000) methods are considered. The use of these “apparently” antagonist methods allows getting a global vision about the applicability of these methods for best estimate tools (Wickett et al., 1998).

Evaluation of the nuclear power plant performances during transient conditions has been the main issue of safety researches since the beginning of the exploitation of nuclear energy. Nowadays, there is a tendency to perform accident analysis by BE coupled codes technique. The main features of such approach is that it allows to get more realistic simulation and more precise consideration of multidimensional effects under normal and abnormal transient conditions in NPPs. The current framework constitutes a contribution for the validation of the coupled code method against the well-documented Kozloduy VVER1000 pump trip experiment through the use of the coupled RELAP5/PARCS code. Through this study, it was revealed that at steady state level, the simulation errors are more probably due to the absence of ADF correction and the used models for evaluating of the Doppler feedback effect. During the transient, the discrepancies are mainly due to the combined effect of uncertain parameters related to the measurement of control rod course, and the estimation of the Doppler effect. The simulation study was complemented by uncertainty and sensitivity analysis in order to assess the error margins and rank the most influent parameters that governs the transient evolution. For these purposes two approaches were applied, i.e. the deterministic CIAU, and respectively, the GRS statistical methods. Differences between the results obtained by the two methods were emphasized as well as their inherent limitations. Additional assessment work should be performed in order to combine the strength of the deterministic and statistical methods and eliminate most of their respective limitations.