Experimental Validation of TRAC-RELAP Advanced Computational Engine (TRACE) for Simplified, Integral, Rapid-Condensation-Driven Transient

Abstract

The purpose of the present work is to experimentally validate the TRACE (TRAC-RELAP Advanced Computational Engine) plug-in of the Nuclear Regulatory Commission's (NRC's) Symbolic Nuclear Analysis Package (SNAP) for rapid condensation transients. These transients are challenging for the code.

The experimental phase began by constructing and calibrating a simplified, integral, condensation-driven transient apparatus named the UMD-USNA Near One-dimensional Transient Experimental Assembly (MANOTEA). Then, a series of 5 well-defined transients were run. Data from the facility included: pressure, differential pressure, and temperature, all as a function of time. Using the data, mass and energy balances were closed for each experiment. Some of the relevant characteristics of the data included: inverted thermal stratification and nozzle dependent transients controlled by an energy partition. A common transient sequence was identified and served as the fundamental comparison to evaluate TRACE.

The second phase began by developing a 1-dimensional Base TRACE Model. Output from the Base Model was found to over-estimate the pressures and temperatures observed in the experiment. This Model always predicted that the condenser pipe would fill, and that transients would terminate with a non-physical discontinuity. In an effort to improve the model, a list of phenomena was generated and then mapped to TRACE parameters. The goal was to find unique ways to capture the energy partition and prevent the condenser from filling. Over 250 TRACE cases were run, and the effective and physically justifiable parameters were incorporated into a 3-dimensional Final TRACE Model. The Final Model incorporated non-condensable gases, which provided a mechanism to terminate the transients smoothly. Replacing the PIPE component with a VESSEL component provided a way to model the energy partition. The Final Model under-predicted trends observed in the experiments. Thus, the two models were able to bracket the experimental data.

Comparing TRACE output to the data led to the conclusion that the code's condensation model is over-stated. TRACE's predecessors were also known to have over-stated condensation models. As a result, TRACE will over-predict condensation-induced fluid motion when modeling several thermal-hydraulic situations important to safe nuclear reactor operation. Future work could focus on developing a NOZZLE component for TRACE, comparing the subsequent output to MANOTEA data and improving the TRACE condensation model.