%HQFKPDUN RI '\QDPLF 6LPXODWLRQ 7RROV IRU 0ROWHQ 6DOW 5HDFWRUV
M. Delpech, S. Dulla, C. Garzenne, J. Kophazi, J. Krepel, C. Lebrun, D. Lecarpentier, F. Mattioda, P. Ravetto, A. Rineiski, M. Schikorr, M. Szieberth
&($ &DGDUDFKH )UDQFH 3ROLWHFQLFR GL 7RULQR ,WDO\ (') 5 ' &ODPDUW )UDQFH %XGDSHVW 8QLYHUVLW\ RI 7HFKQRORJ\ DQG (FRQRPLFV +XQJDU\ )=5 5RVVHQGRUI *HUPDQ\ &156 *UHQREOH )UDQFH (1($ %RORJQD ,WDO\ )=. .DUOVUXKH *HUPDQ\
$EVWUDFW - $ EHQFKPDUN RI VRPH FRGHV GHYHORSHG IRU WKH QHXWURQLF DQDO\VLV RI PROWHQ VDOW UHDFWRUV LV SUHVHQWHG 7KH UHVXOWV RI WKH FDOFXODWLRQV DUH FRPSDUHG WR WKH H[SHULPHQWDO GDWD UHSRUWHG IRU WKH 0ROWHQ 6DOW 5HDFWRU ([SHULPHQW SHUIRUPHG DW 2DN 5LGJH 1DWLRQDO /DERUDWRU\ LQ WKH ÂśV %RWK VWDWLF DQG WLPH GHSHQGHQW VLWXDWLRQV DUH FRQVLGHUHG 7KH SKHQRPHQD FRQQHFWHG WR WKH PRWLRQ RI WKH PXOWLSO\LQJ PDWHULDO DUH LQYHVWLJDWHG 7KH UHVXOWV SUHVHQWHG VKRZ D JRRG DJUHHPHQW DPRQJ WKH FRGHV DQG ZLWK WKH DYDLODEOH H[SHULPHQWDO GDWD WKXV SURYLQJ WKH DGHTXDWHQHVV RI WKH FRPSXWDWLRQDO WRROV IRU WKH VLPXODWLRQ RI PROWHQ VDOW V\VWHPV
I. INTRODUCTION The molten salt reactor concept is attracting a novel interest nowadays for some peculiar features that characterize Àuid-fuel systems. The ¿ssile molten salt Àowing through the core and the primary loop constitutes both the nuclear fuel and the system coolant. This design of these reactors allows also to perform on-line refueling, thus increasing the fuel burn-up, simplifying its management and avoiding the need for fuelelement fabrication. The MOST Project started within the 5-th European Framework Program to assess the state of the art in the ¿eld, in all its different aspects. One of the tasks addresses the physical modelling of a molten-salt reactor, which has to be carefully studied for its speci¿c physical features. In fact, the motion of the fuel through the core and outside in the external circuit introduces physical phenomena which need to be adequately accounted for. The convective Àow has a strong impact in particular on the delayed neutron precursor effect, as they are spatially redistributed within the core, thus causing a reduction of their importance in contributing to the chain reaction. Furthermore, a portion of the delayed emissions takes place outside the core and only the undecayed fraction is re-introduced into the system with a different distribution. As a consequence, the role of delayed neutrons can be signi¿cantly reduced, with important effects on the dynamic response of the system. The Molten Salt Reactor Experiment (MSRE) was de signed and operated at ORNL during the 60’s. MSRE was an 8 MW-th reactor with a molten Àuoride salt circulating at 650 C through a core of graphite bars. The reactor was built to test fuels and materials that could be used in thermal breeder and converter reactors and to provide experience with operation and maintenance.
The core of the reactor was small, with a high neutron leakage, and the graphite blocks, used as moderator, were uncladded and in direct contact with the fuel, a mixture of ¿ssile isotopes with Lithium, Beryllium and Zirconium Àuorides. The heat produced in the core was transferred by the fuel salt to a secondary loop through a heat exchanger, and the heat was then dissipated to the atmosphere through a radiator. The experimental program on MSRE started in 1965 using a molten salt of U235 after three years, a small processing facility was attached to the reactor to remove the original Uranium and a charge of U233 was added to the fuel salt. In the last months of operation a small amount of PuF was added to study the behavior of Plutonium in this kind of system. During the years in operation, several experiments were performed in order to assess the feasibility of a reactor using a Àuid fuel. A great deal of experimental data, under different operating conditions, was made available as a result of the experimental activity. The MOST group has decided to attempt a benchmark exercise among seven participants to assess the adequateness of the computational tools available for the physical evaluation of a Àuid-fuel system, with reference to some of the experimental data from MSRE.
II. DESCRIPTION OF THE BENCHMARKS During the nine-years operation of the MSRE at ORNL an extensive set of experiments was performed, in order to investigate the basic reactor physics parameters associated with the operation of circulating fuel reactors at different power levels. The Âżrst step of this program, widely documented in the reports by ORNL, was the experimental determination of the effects of fuel motion on the reactor kinetic parameters and consequently the kinetic behavior of the reactor at zero-power
condition. During this phase static experiments were conducted to evaluate the reactivity effect of the change of the delayed-neutron precursor distribution due to circulating fuel another measured parameter, important for the dynamic behavior of a mol-ten salt system, is the fraction of delayed neutron lost outside the core region. In this respect experiments were performed to evaluate the effect of temperature on the reactivity of the system, as a change in temperature can cause big changes in the fuel density, affecting the mass balance of the system and the moderating ratio. A variety of dynamic tests were also performed to evaluate experimentally the inherent nuclear stability of the MSRE pump start-up and coast-down transients at constant power were performed, in order to determine the transient effects of fuel Àow-rate changes on reactivity. These tests were performed with both the U235 and the U233 fuels. The most signi¿cant conclusion that stems from the results of the zero-power experimental program was the good agreement of the experimental measurements with the results of the ORNL theoretical models and predictions. The second phase of the experimental program, concerning full-power operation, is not as well documented as the ¿rst one, except for a ”partial load” power transient at very low power, performed in order to determine the characteristics of the heat removal from the MSRE fuel system by natural convection of the primary fuel salt. This transient, widely documented in the reports, is run by increasing the heat-removal rate at the air-radiator, letting the core inlet temperature decrease during a period of 360 minutes. The dynamic behavior was controlled entirely by the inherent feedbacks of the system and the power increased from 0.05% to 4.8% of the nominal power. The experimental program performed at ORNL on molten salt reactors offers then a complete and unique overview on the characteristics of circulating fuel systems, analyzing different operating conditions and studying neutron kinetics, thermalhydraulics and feedback effects. For this reason, these experimental results can be fruitfully employed to test the adequateness of physical models for circulating fuel systems, and has been chosen to benchmark the models and codes of the MOST Project participants. Zero-power results are considered for the benchmark exercise to analyze the neutronic behavior of Àuid fuel systems and reproduce kinetic parameters. The observed values of the system reactivity, delayed-neutron losses and feedback coef¿cients are compared to the results produced with numerical codes. The pump start-up and coast-down transients are reproduced numerically, in order to test the capability of the codes to predict the effects on the dynamics of the change of the salt Àow rate in the system. The transient in natural convection regime allows to validate the performance of the codes accounting for the coupling of neutronics with thermal feed-
back models. III. DESCRIPTION OF CALCULATIONS PERFORMED The aim of the calculations performed during the benchmark exercise has been the validation of physical models and numerical codes used in the MOST Project to study the effects of motion on static parameters and dynamic evolution. Six institutions participated to the benchmarks. They are listed in the following: Budapest University of Technology and Economics (Hungary), BUTE Electricité de France (France), EDF Forschungszentrum Karlsruhe (Germany), FZK, with two participants Forschungszentrum Rossendorf (Germany), FZR Ente per le Nuove Tecnologie, l’Energia e l’Ambiente (Italy), ENEA Politecnico di Torino (Italy), POLITO. The codes employed have different characteristics, so the aim of the exercise is to compare the results and assess the consistency of different models in describing the dynamic behavior of molten salt reactors. The neutronic models implemented in the codes include both transport (deterministic and Monte Carlo) and diffusion solved with different numerical schemes. BUTE uses a modi¿ed version of MCNP4C code, adapted to take into account the transport of delayed neutron precursors. EDF uses a 1D diffusion model with two neutron energy groups. The time inte gration is carried out by a quasi-static approximation. ENEA solves the time-dependent multigroup monodimensional diffusion balance equations, using both full-spatial calculation with an implicit scheme or a modi¿ed point kinetic model. FZK participates with two codes, SIMMER and SimADS: results are identi¿ed in the following with the subscripts (a) and (b), respectively. The SIMMER code solves the 2D/3D neutron transport equation accounting for the axial Àow of the fuel. SimADS code uses a point kinetic model, taking into account the fuel motion in the point equations. The FZR code, DYN1D-MSR, is a one-dimensional nodal two-group neutron diffusion code. The POLITO code solves the time-dependent multigroup 2D diffusion equations, adopting the quasi-static approximation associated to a modi¿ed point kinetic system of equations. The two-group cross sections for the system have been generated by EDF with the APOLLO code, while for the delayed precursor decay constants and fractions different sets of data have been used. In a ¿rst stage the physical data for the families of precursors has been taken from Oak Ridge report then other set of data, produced with JEF database, have been proposed, in order to reduce the discrepancies with MSRE experimental results.
The results obtained with the codes have been compared to the experimental values retrieved in the reports. In the following sections the results of static and dynamic calculations are illustrated.
Table 1. Reactivity lost due to fuel motion in MSRE.
U235 salt
IV. STATIC BENCHMARKS RESULTS The ¿rst part of the benchmark exercise consists in static calculations to evaluate the main physical parameters of circulating fuel reactors and the effects of the motion of the fuel on the reactivity of the system. The results produced with the numerical codes are compared to the experimental values of MSRE (identi¿ed in the tables and graphs as MSRE) and to the results of calculations performed at ORNL (indicated by ORNL). For the Monte Carlo calculation carried out by BUTE the standard deviation for the k and the reactivity is 13 pcm, while for the temperature coef¿cients is 0.184 pcm/C The calculated losses of reactivity due to delayed precursor motion and recirculation are shown in Table 1. All the numerical calculations show a good agreement and consistent behavior, even if they all over-estimate the loss of reactivity. It is supposed that this over-estimation is caused by the uncertainties concerning the geometrical con¿guration of the system to be considered for the calculations. The effect on reactivity of the fuel motion is strongly dependent on the geometry of the upper and lower plena considered, as they are ¿lled with pure ¿ssile salt, hence with a higher multiplicativity than the mixture of salt and graphite that constitutes the core itself. Furthermore, the use of inappropriate delayed neutron data might also give a contribution. Therefore, beside the data provided by ORNL, also more up-to-date JEF data are used. Tables 2 to 5 show the evaluated of the six precursor families for the two types of fuel considered for the benchmark. It can be seen that, as in the previous case, the results of the calculations are larger than the experimental results. Furthermore, some discrepancies can be found comparing values obtained with different codes. These differences are due to the different models used for the evaluation of the loss of importance of the delayed neutrons in presence of fuel motion. This can be calculated performing criticality calculation setting in motion just one group of precursors at a time, in order to separate the effect of each family. Otherwise, the can be evaluated as the difference between the physical value and the effective value of obtained in a point kinetic formulation. In Table 6 the experimental values of the temperature coef¿cients of MSRE are compared to the results obtained in the benchmark exercise. All the results show that the numerical tools of the participants of the MOST Project seem to be adequate to evaluate the main neutronic parameters of a Àuid-fuel reactor such as MSRE.
U233 salt
ORNL data 212 222 244 229 245 262 212 253 261 100 8( 100 85 108 111 125 105 121 119
MSRE ORNL BUTE EDF ENEA FZK (a) FZK (b) FZR POLITO MSRE ORNL BUTE EDF ENEA FZK (a) FZK (b) FZR POLITO
JEF data
260 208 223 258 n.a. 223 236
119 101 104 129 n.a. 112 112
Table 2. Loss in the delayed neutron precursor fractions due to fuel motion in the U235 system ORNL decay data are used. MSRE ORNL BUTE EDF ENEA FZK (a) FZK (b) FZR POLITO
Tot 212 222 244 228.8 259.2 262.2 212.2 253.2 278.0
1
2
3
4
5
6
12.0 14.0 14.1 12.6 13.8 16.0
78.0 90.5 90.8 77.5 89.2 100.7
62.3 71.1 70.9 52.8 68.4 74.1
73.7 80.4 81.9 62.4 77.8 82.8
2.8 3.2 4.1 5.6 3.9 4.0
0.0 0.1 0.3 0.9 0.1 0.2
Table 3. Loss in the delayed neutron precursor fractions due to fuel motion in the U235 system JEF decay data are used. MSRE ORNL BUTE EDF ENEA FZK (a) FZK (b) FZR POLITO
Tot 212 222 224 207.6 234.5 258.3 n.a. 223.0 251.7
1
2
3
4
5
6
12.8 14.9 16.4
65.6 76.2 83.0
55.3 62.9 68.3
71.4 77.9 85.8
2.5 2.9 4.5
0.0 0.0 0.3
10.2 17.0
74.6 84.6
60.5 65.5
75.1 80.4
2.6 4.0
0.0 0.2
V. DYNAMIC BENCHMARKS RESULTS The ¿rst transient considered is the pump start-up. The system starts from a con¿guration with no fuel motion and gets up to the nominal Àow rate in 10 s. During the tranTot 1 2 3 4 5 6 sient the central control rod was withdrawn to maintain critMSRE 100.5 icality and the experimental values used for the benchmark are ORNL 100.5 12.4 43.9 28.2 15.6 0.4 0.1 the positions of the control rod, converted into reactivity usBUTE 85 ing the worth curve available on ORNL report. The reactivity EDF 107.8 12.4 44.1 30.8 20.0 0.5 0.0 inserted into the system during the transient is calculated with ENEA 122.4 14.4 51.4 34.7 21.5 0.5 0.0 the dynamic codes and compared to experimental values. It FZK (a) 125.0 14.5 51.7 35.1 22.7 0.8 0.1 can be seen in Figs. 1 and 2 that all the simulation tools are FZK (b) 105.4 13.6 45.1 26.7 18.4 1.2 0.3 able to follow the physical behavior of the system. In particFZR 121.0 14.3 50.6 33.6 21.2 1.4 0.0 ular, the overshooting of the reactivity, due to the limitation POLITO 134.5 16.7 57.7 36.3 22.9 0.8 0.1 to the velocity of the control rod withdrawing is reproduced by all the calculations. Some difference can be observed in the amplitude of the oscillations due to the re-entering of the undecayed precursors from the external loop. This is probably due Table 5. Loss in the delayed neutron precursor fractions due to to the model of the Àow pattern, which does not take into acfuel motion in the U233 system JEF decay data are used. count multidimensional Àow and its effects on the smoothening of oscillations. Furthermore, the values of reactivity at the Tot 1 2 3 4 5 6 end of the transient show the same over-estimation as in the MSRE 100.5 static benchmark results. ORNL 100.5 12.4 43.9 28.2 15.6 0.4 0.1 The coast-down transient results are presented in Figs. 3 BUTE 131 and 4. In this case the system starts form the critical condition EDF 101.4 12.4 31.7 28.9 27.9 0.5 0.0 in presence of fuel motion, then the Àow rate is stopped in 20 ENEA 114.7 14.5 36.9 33.0 30.0 0.5 0.0 s. As in the previous case, the numerical results reproduce FZK (a) 128.6 16.2 40.8 34.9 35.2 1.4 0.1 the experiment with a good agreement. If all the simulation FZK (b) n.a. results are re-normalized to the experimental loss of reactivity FZR 112.2 14.3 36.2 31.8 29.3 0.6 0.0 of MSRE it can be veri¿ed that the numerical curves reproduce POLITO 126.0 16.8 41.6 34.7 31.8 1.1 0.0 very closely the experimental values. The third experiment considered is the natural convection transient, as described in the previous section. In this case, the neutronic codes are coupled to feedback models and relaTable 6. MSRE temperature coef¿cients [pcm/C]. Effective tions between the jump of temperature across the core and the multiplication constants are evaluated at 650 C. Àow rate are derived. ORNL reports do not give exhaustive and coherent sets of data for the thermal-hydraulics of MSRE, k Salt Graphite Total so different hypotheses on the correlations to be used for the U235 MSRE -8.5 -4.7 -13.2 value of the heat transfer coef¿cient may be made. The refuel ORNL -7.5 -7.3 -14.8 sults of the numerical simulations performed are all in good BUTE 1.0598 -6.2 -4.8 -11.0 agreement with the experimental values, with different levels ENEA 1.07513 -6.8 -4.3 -11.1 of accuracy. In particular, a different behavior at the beginning EDF 1.06752 -7.8 -4.6 -12.4 of the transient can be noticed, due to the fact that the reactor FZR 1.06966 -6.9 -4.0 -10.9 was probably not in an equilibrium state, while steady-state POLITO 1.07060 -6.3 -3.8 -10.1 initial conditions are assumed for the calculations. U233 MSRE -9.5 -5.8 -15.3 fuel ORNL -11.0 -5.8 -16.8 BUTE 1.1271 -8.9 -6.6 -15.5 ENEA 1.13626 -10.6 -5.7 -16.3 EDF 1.12895 -11.0 -5.7 -16.6 FZR 1.12942 -9.9 -4.9 -14.8 POLITO 1.13095 -8.8 -4.5 -13.3 Table 4. Loss in the delayed neutron precursor fractions due to fuel motion in the U233 system ORNL decay data are used.
Fig. 1: Pump start-up transient in MSRE (ORNL decay data).
Fig. 4: Pump coast-down transient in MSRE (JEF decay data).
Fig. 2: Pump start-up transient in MSRE (JEF decay data). Fig. 5: Natural convection transient in MSRE.
VI. CONCLUDING REMARKS
Fig. 3: Pump coast-down transient in MSRE (ORNL decay data).
The MSRE benchmark activities performed during the research performed in connection with the MOST Project have shown that the physical models, numerical tools and nuclear data are adequate to represent the main features of Àuid-fuel systems. Although it has been dif¿cult to ¿nd extensive and consistent sets of results for experiments performed 40 years ago, the MSRE available data have been reproduced with good agreement. The benchmark activity has been also very fruitful in deepening the understanding of the basic physics of moltensalt systems and in establishing a good collaboration relationship among all the participating Institutions.
VII. REFERENCES 1. B. E. PRINCE, J. R. ENGEL, S. J. BALL, P. N. HAUBENREICH, T. W. KERLIN, � Zero-Power Physics Experiments on Molten-Salt Reactor Experiment� , ORNL-4233, Oak Ridge, 1968. 2. P. N. HAUBENREICH, � Molten-Salt Reactor Experiments� , ORNL-4396, 1970. 3. J. KOPHAZI, M. SZIEBERTH, S. FEHER, GY. CSOM, P. F. A. DE LEEGE, � MCNP Based Calculation of Reactivity Loss in Circulating Fuel Reactors� , Proceedings of the International Conference on Nuclear Mathematical and Computational Sciences: A Century in Review-A Century Anew, Gatlinburg, April 6-10, 2003. 4. D. LECARPENTIER, V. CARPENTIER, � A Neutronic Program for Critical and Nonequilibrium Study of Mobile Reactors: the Cinsf1D Code� , Nuclear Science and Engineering, , 2003. 5. S. KONDO, K. MORITA, Y. TOBITA, N. SHIRAKAWA, � SIMMER-III: an Advanced Computer Program for LMFBR Severe Accident Analysis� , ANP’ 92, Tokyo, 1992. 6. G. LAPENTA, F. MATTIODA, P. RAVETTO, � Point Kinetic Model for Fluid Fuel Systems� , Annals of Nuclear Energy, , 1759-1772, 2001. 7. F. MATTIODA, P. RAVETTO, G. RITTER, � Effective Delayed Neutron Fraction for Fluid-Fuel Systems� , Annals of Nuclear Energy, , 1523-1532, 2000. 8. S. DULLA, M.M. ROSTAGNO, P. RAVETTO, � QuasiStatic Method for the Time-Dependent Neutronics of FluidFuel Systems� , American Nuclear Society Winter Meeting, New Orleans, 16-20 November, 2003. 9. R. SANCHEZ et Al., � APOLLO II: a User-Oriented, Portable, Modular Code for Multigroup Transport Assembly Calculations� , Proceedings of the Congress on Advances in Reactor Physics, Mathematics and Computation, Paris, , 15631577, 1987.