Turbulent heat mixing of a heavy liquid metal flow in the MEGAPIE target geometry- The heated jet experiment

Full text of publication follows: The MEGAPIE target to be installed at the Paul-Scherrer Institute in Switzerland is a prominent example of a window spallation target using liquid lead-bismuth both as coolant and neutron source. An adequate cooling of the target to maintain the temperatures in the structure within acceptable limits requires a conditioning of the flow. In MEGAPIE this is realized by a main flow transported downwards, u-turned at the proton beam facing hemispherical shell into a cylindrical riser tube. In order to avoid a stagnation point close to the lowest part of the shell a jet flow is directed along the shell, which is superimposed to the main flow. The heated jet experiment conducted in the THESYS loop of the KALLA laboratory is a 1:1 representation of the lower part of the MEGAPIE target and is aimed to study the cooling capability of the specific geometry. In this out-of pile experiment a heat jet is injected into a cold main flow at MEGAPIE relevant flow rate ratios. The resulting temperatures are recorded in dense instrumented critical part of the lower shell region. The experiment is accompanied by a 3D turbulent numerical simulation. Besides the nominal operation point deviations from it are studied by means of varying main or jet flow rate in order to give an estimate on the stable operation limits. The experiment demonstrated that any of the flow rate configurations studied revealed a time dependent behavior in the vicinity of shell region, which could not be reproduced by the CFD, where always a steady solution is obtained. The normalized temperature fluctuation intensities in the window region are of the order of 1, but their time scale is rather fast, so that for MEGAPIE hardly any temperature fluctuations of the window material can be expected. Both the temperatures at the fluid wall interface of the window and the temperature mixing in the riser tube are significantly underestimated by the simulation, so that the CFD can be considered with regard to this issue as a conservative assessment. More serious in this context is that the numerical qualitatively fails to describe the temperature distribution, which is a motivation to the development for liquid metal adapted turbulent heat transfer models. The temporal analysis of the thermocouples exhibited a complex multi-stream and multi-vortex pattern of the fluid flow in the shell region, which depends only on the flow rate ratio of main to jet flow and not to their absolute values, the jet inlet temperature etc. Moreover, the spectral density of the temperature readings confirmed that the turbulence in the shell region and the riser is mainly isotropic and governed by inertia effects. For flow rate ratios of main to jet flow larger 12.5 the jet splits into several streams as it hits the shell. One of the sub-streams is re-hitting the shell again. The evolving pattern is rather instable and normal to the nozzle plane at the centerline an insufficient cooling is obtained. This effect increases as the flow rate ratio grows up to values of 20 and more. For ratios less than 12.5 the jet covers the whole shell both in the nozzle plane and perpendicular to it, which leads to a sufficient window cooling. As the flow rate ratio falls below 11.1 the jet even enters the gap duct opposite the nozzle, which leads to larger fluctuation levels in the shell region, which are accompanied by undesired larger pulsation times. Nevertheless, the pulsation times with little less than 1 Hz are acceptable in terms of a safe MEGAPIE operation. (authors)