One case recently completed to validate the three-dimensional hydrodynamics model inRELAP5-3D^© using test data is a simulation of LOFT large breakloss-of-coolant-experiment (LOCE) L2-5¹³. The LOFT facility was a 50 MWpressurized water reactor (PWR) that was designed to simulate the response of a commercialPWR during a loss-of-coolant accident (LOCA). Test L2-5 simulated a double-ended offsetshear of a cold leg. The experiment was selected because it was judged to provide the mostchallenging test of the multi-dimensional model of all the experiments in the existingdevelopmental assessment test matrix used for RELAP5/MOD3.

The analysis was accomplished in several steps. First, the original one-dimensionalmodel was upgraded to be consistent with current user guidelines and to better representTest L2-5. The upgraded one-dimensional model was then run with the current version ofRELAP5-3D^© to obtain baseline results.

The RELAP5/MOD3 model that was used in the developmental assessment of LOFT Test L2-5is illustrated in Figure 11. The model, hereafter referred to as the one-dimensionalmodel, contains 131 volumes, 142 junctions, and 77 heat structures.

The three-dimensional model of the LOFT reactor vessel was developed using twomulti-dimensional components. Multidimensional component 700 represented the downcomerregion as shown in Figure 12. Multidimensional component 200 represented the lower plenum,core inlet, core, and upper plenum regions as shown in Figure 13. The vessel was dividedinto four 90^o azimuthal sectors and four radial rings. The four azimuthalsectors corresponded to the four nozzles connecting the loops and the vessel. One radialring (multidimensional component 700) represented the downcomer while the other threerings (multidimensional component 200) corresponded to high-, average-, and low-poweredregions of the core. The axial nodalization of each multi-dimensional component was basedon that of the one-dimensional model, resulting in 6 levels for the downcomer and 21levels for the lower plenum, core inlet, core, and upper plenum regions. A multiplejunction (Component 709) connected the bottom of the downcomer to the top of thethird ring in the lower plenum. The three-dimensional model of the reactor vessel was theninserted into the one-dimensional model, with the resulting model hereafter referred to asthe three-dimensional model.

The core fuel rods were modeled with twelve heat structure geometries, eachrepresenting the fuel rods located in a given ring and sector. Each fuel rod heatstructure geometry contained twelve heat structures, corresponding to the number of axiallevels in the hydraulic nodalization of the core. Each fuel rod heat structure geometrywas identical except for its hydraulic connections and the radial power peaking factor.The radial power peaking factors were 1.31, 1.13, and 0.81 for the inner, middle, andouter rings respectively. For comparison, the radial power peaking factors for thehigh-powered and low-powered rods in the one-dimensional model were 1.31 and 0.94,respectively. The maximum axial power peaking factor was 1.58 and was located in thebottom third of the core.

Figure 13. Nodalization of the 3D Model of the LOFTLower Plenum, Core Inlet, Core, and Upper Plenum Regions

Steady-state calculations were performed for LOFT Test L2-5 with both theone-dimensional and three-dimensional models. Table 1 shows that the results of thesteady-state calculations were in excellent agreement with the measurements.

Transient simulations of LOFT Test L2-5 were performed using the one-dimensional andthree-dimensional models. Comparisons between calculated and measured results, includingthe sequence of events, the overall system response, and the core thermal response, arenext discussed.

The measured sequence of events for LOFT Test L2-5 is presented in Table 2. Thetest was initiated at 0.0 s when the quick-opening blowdown valves began to open. Areactor trip signal was generated at 0.02 s on low hot leg pressure, and the reactorscrammed shortly thereafter. The operators tripped the primary coolant pumps at 0.94 s.Flow from the accumulator, HPIS, and LPIS began at 16.8 s, 23.90 s, and 37.32 s,respectively. The accumulator emptied at 49.6 s. The fuel rod cladding temperature firstdeparted from near the saturation temperature at 0.91 s. The peak cladding temperatureoccurred at 28.47 s. All of the core cladding had been quenched by 65 s, concluding theinteresting portion of the test.

Table 2 also presents the calculated event times with the one-dimensional andthree-dimensional RELAP5 models. The calculated event times with both RELAP5 models weregenerally in reasonable quantitative agreement with the measured values. One exception wasthat the calculated peak cladding temperatures with both models occurred near 6 s,compared to about 28 s in the test. As will be shown later, the measured claddingtemperatures increased slowly between 5 s and 28 s while the calculatedtemperatures decreased slowly, so the effect of the difference in timing is not assignificant as might be inferred from Table 2. The core cladding also quenched 10 sto 15 s earlier in the calculations than in the test.

The results of the assessment calculations are presented in the form of comparisonplots, which show measured values and the corresponding calculations with theone-dimensional and three-dimensional RELAP5 models.

A comparison of calculated and measured primary system pressures is presented in Figure14. The calculated primary system pressures were similar with both models and inreasonable agreement with the data. The measured curve contains an inflection point atabout 16 s, roughly corresponding to the initiation of accumulator injection that was morepronounced than in the calculations. The calculated pressures were also slightly less thanthe measured values after 40 s.

Table 2. Calculated and measured sequence of eventsfor LOFT Test L2-5.

1. The measured voltage and current to the pumps did not drop instantaneously to zero following the trip of the pumps in the test. Since the code assumes an instantaneous trip, the pump trip in the calculations was delayed to coincide with the measured decrease in pump speed and differential pressure.

Figures 15 and 16 show calculated and measured mass flow rates in the broken loop coldleg and broken loop hot leg. The measured flow in the broken loop cold leg wassubstantially larger than the flow in the broken loop hot leg, particularly during thefirst 5 s of the transient. The fluid upstream of the cold leg break was subcooled forseveral seconds while the fluid upstream of the hot leg break was at the saturationtemperature almost immediately after the break, leading to higher critical flow rates onthe cold leg side. Both RELAP5 models predicted this trend.

The calculated results are generally within the uncertainty of the measurements and arethus considered to be in excellent agreement with the data.

A comparison of calculated and measured mass flow rates in the intact loop hot leg isshown in Figure 17.

The instrument measured only the magnitude of the flow rate and not its direction.Consequently, the absolute values of the flow rates, which are presented inFigure 17, provide a more direct indication of the agreement between the calculatedand measured results. In the calculations, the flow in the hot leg was generally towardsthe steam generator until 5 s. The flow then reversed, going towards the reactor vesseldue to the pump trip and the corresponding flow coastdown. The maximum negative flowoccurred at about 10 s and was caused by vapor generation in the steam generator u-tubes,which forced flow towards the reactor vessel. The draining of the pressurizer alsocontributed to the flow from the hot leg to the reactor vessel. Based on the comparisonsshown in Figure 17, a similar flow reversal probably occurred in the experiment. Thetrends in the calculations were similar to those observed in the test except that themeasured results were more oscillatory, particularly between 35 s and 60 s, which roughlycorresponded to the reflooding of the core.

In general, there was little difference between the results from the one-dimensionaland three dimensional models insofar as loop behavior was concerned. That was notunexpected. Differences were obviously expected in the vessel.

Figure 18 presents calculated and measured fuel centerline temperatures in the central,high-powered bundle. The measured temperature decreased rapidly during the first 5 s ofthe transient, then remained nearly constant until 63 s, when the temperaturedecreased rapidly to near the saturation temperature of the fluid. The rapid temperaturedecrease at 63 s is referred to as a quench and indicates that the fuel rod surface hadbeen wetted by the advancing mixture level during the reflooding of the core. Thecalculated temperatures were similar to the measurement except that the quench occurredearlier than in the test, particularly in the calculation with the one-dimensional model,and both calculated temperatures decreased at a faster rate than in the test between 5 sand the quench time.

Comparisons between calculated and measured cladding temperatures as a function ofelevation are presented in Figures 19, 20, and 21. The results correspond to axial levels6, 7, and 8 of ring 1, sector 2 of the three-dimensional model. Ring 1 represents most ofthe central, high-powered fuel rod bundle, and sector 2 represents the quadrant connectedto the broken loop hot leg. The one- and three-dimensional models produced similarresults, underpredicting the peak cladding temperature and quenching earlier than in theexperiment. However, quenching behavior in the three-dimensional model more closelymatched the data. This is mainly attributed to the fact that the high-powered fuel rodswere attached to a relatively hot fluid channel in the three-dimensional model whereas thehigh-powered fuel rods were attached to a single, average channel in the one-dimensionalmodel.

Significant radial variations in cladding temperatures were observed in both the testand the calculation with the three-dimensional model as shown in Figures 22 and 23.Figure 22 shows measured temperatures in rings 1 through 3 at level 9. Thethermocouples referred to in the figures were all located at the same elevation and withinthe same sector so that the measured differences in results were due to radial effects.Figure 23 shows the corresponding calculated results with the three-dimensional model. Thecalculated and measured temperatures both show the influence of the radial power profile,with the highest temperatures in ring 1, the high-powered ring, and the lowesttemperatures in ring 3, the low-powered ring. The radial variation in claddingtemperatures was more pronounced in the test than in the calculation, primarily becausethe time of CHF varied more in the test.

The hydraulic responses of the primary and secondary coolant systems calculated withRELAP5-3D^© and the three-dimensional input model weregenerally in reasonable agreement with LOFT Test L2-5. The results were generally as goodas or better than the results obtained using the RELAP5/MOD3 or RELAP5-3D^© one-dimensional models. The calculated thermal response of thecore fuel rods with the three-dimensional model was also generally similar to thatobserved in the test. The calculated peak cladding temperature was 990 K while themeasured peak cladding temperature was 1078 K. The most significant deviations between thecalculated and measured thermal responses were that the calculated peak claddingtemperature occurred earlier than in the test and that the top-down rewet that wasobserved near 15 s in the test was not predicted.

Part 3 of the detailed paper discusses the mutlidimensional neutron kinetics model andBorder Profiled Lower Upper (BPLU) sparse matrix solution technique implemented inRELAP5-3D^©.  This link will take you to Part 3 of thedetailed paper. You may also return to Part 1 of thedetailed paper or the brief paper.

Questions or problems regarding this web site should be directed to danidl@inel.gov.

Last modified: Thursday November 12, 1998.