Uncertainty and sensitivity analysis of water hammer phenomenon by employing the Umsicht test facility data

Full text of publication follows: Pressure surges occurring in pipeline systems may be caused by fast control interference, start up and shut down processes and operation failure, as well as flow rate fluctuation. They lead to water hammer upstream the closing valve and cavitation hammer downstream the valve, which may cause considerable damages to the pipeline and the support structures. Appearance of water hammer in thermal-hydraulic systems was widely studied employing different state-of-the-art thermal-hydraulic codes in many organizations. For the analysis we consider water hammer test performed at Fraunhofer Institute for Environmental, Safety and Energy Technology (UMSICHT) at Oberhausen. This paper presents the comparison of UMSICHT test facility experiment calculations employing the best estimate system code RELAP5/Mod3.3 to measured at UMSICHT test facility water hammer values at various tap water velocities after fast closure of a valve. The analysis revealed that RELAP5/Mod3.3 code calculated the first pressure peak matches the measured value of pressure very well, but concerning the following pressure peaks, the measured pressures are higher and the peaks appear with an accumulating time delay. However, the first cavitation hammer because of achieved the peak pressure value is the most dangerous and can lead to damages of plant's equipment (valves, pumps, pipe bends) up to leakage of the pipe system. These damages have direct impact on plant availability and plant safety therefore water hammer effects investigation is important at safety justification of NPPs. Results of each individual calculation always contain uncertainty owing to initial conditions of installations, errors of measuring systems, errors caused by nodalization of objects at modeling, code correlations, etc. In this connection an actual task is the evaluation of calculations results uncertainty. In the paper, results of uncertainty and sensitivity analysis of the initial conditions and code-selected models are shown. Such analysis allows to estimate the influence of separate parameters on the calculation results and find those modeling parameters that have largest impact on the first cavitation hammer peak. On the basis of this analysis, recommendations for further improvement of model are developed. (authors)