Monte-Carlo/Simmer-III Reactivity Coefficients Calculations for the SuperCritical Water Fast Reactor Magnus Mori1, Werner Maschek1, Eckart Laurien2, Koji Morita3 1
Forschungszentrum Karlsruhe, Institut für Kern- und Energietechnik (FZK/IKET) 76344 Eggenstein-Leopoldshafen, Germany
[email protected] 2 Institut für Kernenergetik und Energiesysteme (IKE), Universität Stuttgart Pfaffenwaldring 31, 70569 Stuttgart, Germany 3 Institute of Environmental Systems, Graduate School of Engineering, Kyushu University 6-10-1 Hakozaki, Higashi-ku, Fukuoka 812-8581, Japan ABSTRACT – In recent years more and more attention was dedicated to the development and detailed analysis of supercritical water cooled reactors [1]. In particular several studies (e.g.: [2,4]) outlined that the SuperCritical water cooled Fast Reactor (SCFR) presents new features that require a careful analysis of its neutronics safety parameters. Specifically it is considered of the highest importance to assess the void effect accurately, given the fact that a negative void reactivity is a fundamental requirement for the inherent safety of water cooled reactors; since, differently from a sodium cooled reactor, Loss Of Coolant Accidents (LOCAs) are a Design Basis Accident (DBA) for SuperCritical Water cooled Reactors (SCWR). With the aim of accurately estimating the void reactivity coefficient several models were then developed and employed pursuing the objectives of improving the void effect to overcome the computational uncertainties [3] and investigating the possibility of performing coupled neutronics/thermal-hydraulics calculations using MCNP4C and SIMMER-III. The paper will then first illustrate the outcome of the initial void reactivity effect investigations and then the details of the advanced models that were developed to analyze the problem more accurately. Concluding then design modifications are suggested and the results relative to the previously mentioned coupling are discussed. I. INTRODUCTION New nuclear reactors are currently designed not only for the production of electricity, but also with the aim to employ them as effective transmuters, radioactive waste burners, and, therefore, with more attention to the overall fuel cycle and treatment of the waste. In this context the SuperCritical water cooled Fast Reactor (SCFR) is investigated for its potential as a transmuter for minor actinides. The SCFR is a once-through direct cycle reactor [1], where all feed water flows through the core to the turbine at supercritical pressure. It is characterized by an epithermal/fast neutron spectrum, which would allow its employment as a transmuter. In the present context of nuclear research this feature represents a most valuable incentive for the further development of this reactor. Furthermore there could be a potential to avoid the build up of curium in the fuel inventory, which is a typical drawback of minor actinides (MAs) burning in light water reactors (LWRs). A conversion ratio greater than 1 is then considered to be an important characteristic, since it could eventually allow burning spent fuel in dedicated targets that could exploit a specific feature of the SCFR namely the fast spectrum and thermal spectrum region existing in
Global 2003
New Orleans, LA
November 16-20, 2003
the seed/blanket parts of the core to achieve an optimum conversion of the MAs. One of the requirements for the achievement of a fast spectrum is a tight lattice configuration [2], which is one of the specific characteristics of this design that, together with the highly enriched fuel composition (~23%) and eventually with a relatively large content fraction of minor actinides, arises questions regarding the safe behavior of the reactor under transient conditions. The safety parameters of a supercritical water cooled fast reactor are therefore investigated to assess the basic feasibility of such a design adopting a refined MCNP [7] core model for the neutronics analyses and SIMMER-III [8] with extended supercritical water equations of state for the fluid dynamics, specifically focusing on the void reactivity coefficient. Furthermore the effect on the void effect coefficient of various core rearrangements that account for different designs of the blanket/seed regions is another item of this study. II. VOID EFFECT ESTIMATIONS FOR THE SCFR A detailed void effect analysis using different cross section data libraries (ENDF/B-VI.7, JEF-2.2 and JENDL-3.2) and different models was performed as described in detail in [3]. Because of the pronounced 1754
implemented: improving the void reactivity coefficient, and improving the adopted models, hence reducing the uncertainties. More calculations were then performed using the most up-to-date versions of the available cross section data libraries (ENDF/B-VI.8, JEFF-3.0, and JENDL-3.3). Upper Axial Blanket
24.23 (14.04)
22.07 (12.79)
23.37 (13.54)
Blanket
22.07 (12.79)
22.93 (13.29)
Outer Radial
23.80 (13.79)
21.20 (12.28)
Blanket 2
24.23 (14.04)
22.93 (13.29)
Central Radial
Blanket
A mid plane axial section of the SCFR core is shown in Fig. 1, which gives an idea of the mentioned highly heterogeneous structure of this reactor (the light blue elements represent the radial blanket regions, within which dark blue elements can be distinguished representing the solid moderator pins, the red areas represent the seed. The seed areas are made of plutonium enriched uranium (see Fig. 2), the blanket areas of depleted uranium, and the solid moderator pins of zirconium hydride.
Inner Radial
II.A. Description of the SCFR core
21.64 (12.54)
Blanket 1
24.23 (14.04)
Central Radial
heterogeneity of the core [4] several full-scale MCNP models (both 2D and 3D) were then developed and applied with diverse nuclear data libraries and various approximations in the implementation of these data (related to various approaches of taking into account selfshielding effects in the unresolved resonance energy region, delayed neutron spectra, and thermal scattering data). In order to better understand the importance of these analyses and models a short description of the core will be given hereafter.
23.80 (13.79)
Lower Axial Blanket
Fig. 2 Core radial section and Pu enrichment distribution (weight % of 239Pu and 241Pu in parentheses) TABLE I. Void effect variation with different approximations in nuclear data treatment. Model Approximations No Prompt Free gas Free gas probability fission H in H in tables spectrum ZrH water Total void reactivity variation Min/Max ($) -0.03 -0.62 -0.04 -1.12 /+0.93 /+0.71 /+0.33 /-0.68
Fig. 1. Core axial section. III. REFINED VOID EFFECT ANALYSES Because of the uncertainties in the void reactivity coefficient estimation (see Table I) it was concluded that in order to guarantee a negative void effect, a conservative value of –5$ (1$ = ~377pcm for the examined core configuration) should be predicted. Two different, but closely related, strategies were then
Global 2003
New Orleans, LA
November 16-20, 2003
Part of the refinement work concentrated on the MCNP model. The introduction of a detailed 3D geometry showed the importance of the neutron streaming effect for this design (for the ENDFB-VI.7 data file, which proved to be the most conservative cross section library for this design, the void reactivity coefficient shifted from ~+1$ to ~-0.5$). Consequently, several sensitivity studies were performed on subassembly gap spacing, radial and axial blanket geometries, and solid moderator hydrogen enrichment. Another consistent part of the refinement work focused on the introduction of fuel temperature and coolant density profiles. III.A. Description of the new advanced MCNP model The main goal of the new model is to take into account the spacing among the sub-assemblies, a correct three-dimensional description of the upper, lower and radial reflector, and of the axial blankets. Furthermore, it shall allow for more flexibility to define the physical
1755
IV. REFINED ANALYSES: RESULTS One of the main known drawbacks of MCNP is the so-called cell approximation. The absence of continuous temperature treatment and density feedbacks implies that the functions describing these parameters need to be defined stepwise. This introduces an obvious error that can be judged by looking at the following figures (Fig. 3 and Fig. 4), which show the temperature and density profile cell approximation for the SCFR.
water and zirconium hydride thermal scattering, an average fuel temperature 1200K, and an average coolant temperature of 800K). T coolant
rho coolant
900
850 compressed
supercritical
800
800
700 750
600
700
500
650
400
rho [kg/m3]
T [K]
properties of different regions of the reactor. Although the model adopted in [3] was based on a single pin discretization, the pins were not described individually, but rather reproduced several times retaining their original properties, and assigning a different position making use of the universe definition feature of MCNP. A characteristic of relevance that was observed when investigating the effect of the introduction of control rods, was the high axial decoupling of the SCFR core. This pointed out the necessity of taking into account not only the correct geometrical description but also the appropriate density and temperature profiles of the materials used in the core, and in particular of the coolant. It was therefore decided to increase the number of universes included in the calculations, improving the detail of the description of the problem. In the new model then not only different universes were created to describe the different geometries, but also to take into account the different properties of the materials e.g.: densities and temperatures, since as known MCNP cannot take into account continuous distributions of physical properties.
300
600
200 550
100
500
0 0
0.5
1
1.5
2
2.5
3
core height [m]
Fig. 4. 16 cell coolant density & temperature profiles. TABLE II. Void reactivity effect variation with detail of coolant axial profile. 1 mesh
[$]
gap=0 mm
k-eff
+/-
[pcm]
nominal
0.98017
62
0
full void
0.97076
76
-989
-2.7
gap=0 mm
k-eff
+/-
[pcm]
[$]
nominal
0.98276
57
0
full void
0.97039
56
-1297
-3.5
gap=0 mm
k-eff
+/-
[pcm]
[$]
nominal
0.98281
58
0
full void
0.97105
51
-1232
-3.3
gap=0 mm
k-eff
+/-
[pcm]
[$]
nominal
0.98212
57
0
full void
0.97113
44
-1152
4 meshes
8 meshes T coolant
rho coolant 800
800 compressed
supercritical
700
750
600 500 400
650
300
600
16 meshes rho [kg/m3]
T [K]
700
-3.1
200 550
100
500
0 0
0.5
1
1.5
2
2.5
3
core height [m]
Fig. 3. 4 cell coolant density & temperature profiles. Given the difficulty to decide a-priori the ideal number of cells, 4 calculations were performed considering 1, 4, 8, and 16 axial cells. The profiles were calculated assuming an average cosine shaped axial power distribution. The results are summarized in Table II (these calculations were performed with ENDF/B-VI.7,
Global 2003
New Orleans, LA
November 16-20, 2003
The 16 mesh input deck was taken as the reference model since k-effective is not changing too significantly once an axial profiles, even a rough one, is introduced. This analysis, however, shows that a single axial mesh description introduces an error of at least 0.4$, indicating the importance of the introduction of this modeling feature and, moreover, the relevance of the SIMMERIII/MCNP4C coupling work that is being performed. IV.A. Effect of Subassembly Gap Spacing Another geometrical detail that was neglected in the model adopted in [3] is the spacing among the
1756
subassemblies. The gap distance indicated in [5] is 1mm. In view of the fact that the mechanical tolerance required to achieve this value could result being too strict, calculations were also performed for a 2mm gap [6] (the computational options used are the same as described in Table II and 16 coolant density axial profiles were used). Different concurring phenomena contribute to the higher or lower efficiency of the analyzed geometries. The introduction of a gap is expected to affect negatively the k-effective for the nominal conditions because of the introduction of more water, which acts more as an absorber rather than as moderator. The k-effective of the SCFR core under voided conditions decreases with increasing gap spacing, because of a higher leakage fraction. However this effect is counteracted by a larger reduction of the nominal k-effective, which therefore results in a less negative void effect for the 2mm gap case. The 1mm geometry gives a better void positively combining the two mentioned effects.
seed areas from the neutrons was studied. In Fig. 5, it is possible to observe the configuration of a tight blanket lattice arrangement: the spacing among the seed subassemblies and among the seed and the blanket subassemblies is kept at the design value (1mm or 2mm), while the blanket sub-assemblies are tightly arranged adopting a thicker can wall. This solution should prove viable from the construction point of view since it would allow for thermal expansions in the radial and in the axial directions. However the effect of longitudinal expansion could jeopardize this specific core configuration. The engineering feasibility of this core design needs hence further investigation, especially considering the fact that the results relative to this configuration are very promising as shown in Table IV.
TABLE III. Void reactivity effect variation with increasing subassembly gap spacing. gap=0 mm
k-eff
+/-
[pcm]
[$]
nominal
0.98240
64
0
full void
0.96998
49
-1303
-3.5
gap=1 mm
k-eff
+/-
[pcm]
[$]
nominal
0.98100
51
0
full void
0.97064
47
-1088
-2.9
gap=2 mm
k-eff
+/-
[pcm]
[$]
nominal
0.97894
59
0
full void
0.96937
53
-1008
-2.7
IV.B. Enhanced Void Effect Blanket The results obtained with the new MCNP model that includes reflectors, subassembly spacing, coolant density axial profiles, and detailed axial blanket description, prove that the SCFR core has a potentially negative void coefficient. It was argued in [3] and [9], that, all uncertainties taken into account, a negative void of –5$ should be calculated in order to guarantee a negative void in all conditions, assuming a BOC (Beginning Of Cycle) fuel composition. Therefore, with the aim of improving further the void reactivity coefficient a new blanket geometry was investigated. The introduction of reasonable subassembly gap spacing did not improve the void effect. This pointed out that in a voided reactor fast neutrons that stream through the gap spacing contribute significantly to the reactivity level of the reactor “seeing” more seed zones, that is the relative number of neutrons giving fission is higher than the leakages. Having this fact in mind, a tight blanket configuration that could increase the number of captures shielding the
Global 2003
New Orleans, LA
November 16-20, 2003
Fig. 5. Enhanced void blanket configuration. TABLE IV. Void reactivity effect estimation for tight blanket lattice design (see Table III). gap=0 mm
k-eff
+/-
[pcm]
[$]
nominal
0.98212
57
0
full void
0.97113
44
-1152
-3.1
gap=1 mm
k-eff
+/-
[pcm]
[$]
nominal
0.98036
64
0
full void
0.96582
47
-1536
-4.2
gap=2 mm
k-eff
+/-
[pcm]
[$]
nominal
0.97808
53
0
full void
0.96149
54
-1764
-4.8
The subassembly arrangement with no gap spacing does not show significant void effect variations as to be
1757
expected. The difference of 0.4$ can be attributed mostly to statistical uncertainties. The situation changes significantly with larger gap sizes. The void reactivity worth improves by -1.3$ with a 1mm gap and by -2.1$ with a 2mm gap. Nominal k-effectives are lower for the tight lattice configurations because of the high capture cross section of nickel that is replacing water around the blankets. Voided condition k-effective values are also lower because of the combined effect of captures in the can wall, and more captures in the blankets due to a reduction of radial streaming. Hence the void effect for a tight blanket configuration and a 2mm subassembly gap spacing is for –4.8$ ENDFB-VI.7, close to the target value of –5$. Further improvements of the void effect would be achieved by removing the lower axial blanket. Such a solution is commonly suggested to prevent recriticalities in case of reconfiguration of fissile material in the core. Furthermore, the introduction of ZrH2 replacing ZrH1.7 as a solid moderator will improve the void effect. The combined effect of these two design changes would have the effect of making the void effect even more negative achieving a value of –5.8$ IV.C. Application of New Cross Section Libraries In previously published work [3] it was argued how ENDF/B proves to be the most conservative library, if compared to JEF, or JENDL, for SCFR neutronics analysis. For this reason all the new calculations were performed with ENDF/B-VI.7. Data in different nuclear data libraries may vary significantly. Each evaluation may have advantages and disadvantages, the differences reflecting evaluator’s strategies in attributing priorities to different experimental results and employing different fitting techniques. Therefore, results obtained with different libraries may reflect to a certain degree the existing uncertainties in nuclear data. However, the uncertainty of the void effect cannot be derived in a straightforward manner from the computations with ENDF, JEF, and JENDL: in many cases the latest sophisticated experiments have the highest level of credibility. The adoption of newer data libraries should thus decrease the uncertainties inherent with the evaluation of the library itself (e.g.: improvement of the experimental techniques, of the fitting techniques, availability of new data), but does not cancel the differences among the different strategies followed by different groups of evaluators. Table V shows a summary of void reactivity effect estimations performed with the available cross section files. The difference of the void reactivity coefficient between the older and the newer cross section file for the same library evaluation group are approximately in the range of the statistical uncertainty of the Monte-Carlo calculations. It is worth outlining the fact that for the Global 2003
New Orleans, LA
November 16-20, 2003
application here of interest the new release of ENDF/B does not seem to have a significant influence on the absolute results (k-effective), while more appreciable discrepancies can be noticed for both JEF and JENDL. A close examination of the cross section files for 239Pu showed for these two libraries appreciable differences. In particular what strikes most is the lower setting of the fission cross section file of JEFF-3.0 with respect to JEF2.2, which helps to explain the lower k-effective estimated for both nominal and voided reactor conditions. TABLE V. Effect of the new cross section files (compare to Table III).
nominal full void
ENDF/B-VI.7 k-eff [$] 0.97808 0.96149 -4.8
ENDF/B-VI.8 k-eff [$] 0.97783 0.96195 -4.6
nominal full void
JENDL-3.2 k-eff [$] 0.97151 0.95250 -5.6
JENDL-3.3 k-eff [$] 0.97529 0.95357 -6.3
nominal full void
JEF-2.2 k-eff [$] 0.96778 0.94726 -6.0
JEFF-3.0 k-eff [$] 0.97354 0.95167 -6.4
Examining the influence on void reactivity effect of the new libraries it is possible to appreciate how JEFF and JENDL give very close estimates. The biggest differences, in the order of ~2$, can be observed between JEF and ENDF/B, and consequently between JENDL and ENDF/B. Having a close look at the evaluated keffectives it is worth noting the coherence in the evaluation of the nominal k-effective for the new cross section files. For nominal conditions the calculated difference for the new cross section libraries among the estimated k-effective is approximately half of the value calculated with the old files (in the range of 0.5÷1.2$). For voided conditions the difference is comparable to the old values except for JEFF-3.0 vs. JENDL-3.3, for which the difference is about 5 times smaller and comparable to the one relative to the nominal conditions (~0.5$). Concluding then ENDF/B-VI.8 gives the most conservative results for the specific composition and spectrum of interest of the SCFR, as it did ENDF/B-VI.7. V. ENHANCED VOID CORE DESIGN The strategy adopted for the new SCFR core designs is based on the introduction of more heterogeneity and more solid moderator (ZrH1.7). The overall number of blanket/seed sub-assemblies and their ratio is kept constant in order to maintain the same core inventory and approximately the same breeding/burning capabilities. Different configurations were examined and calculations were performed applying the two available MCNP 1758
models: the 3D model illustrated in [3] and [9] and the newest model described and analyzed in this work. The configuration that was chosen is shown in Fig. 6 and is believed it should help increasing the heterogeneity effects improving the void worth. The arrangement of the blanket “petals” is such that a periphery peaked flux distribution results in this core, which in case of voided conditions enhances radial leakage towards the exterior of the core and towards the central shielded blanket areas, increasing the number of captures. The results of the calculations relative to this geometry are given in Table VI (the same model used for the results reported in Table III was applied). The combined effect of the core rearrangement and of the introduction of more precise details in the model did not improve the void reactivity effect significantly. Then again, as illustrated before (Table III) the introduction of a larger sub-assembly gap (2mm) deteriorates the void effect of this reactor.
A second set of calculations was therefore performed applying the tight blanket concept to the just illustrated “flower” model. The results are shown in Table VII and prove once more the importance of the shielding/capturing effect of the blankets for the behavior under voided conditions of this core. Consequently, introducing a 2mm sub-assembly gap spacing together with the central blanket for the advanced heterogeneous model, a void reactivity worth of –5.8$ was calculated with ENDF/B-VI.7, even though the lower blanket was not removed. Another advantage of this configuration would be the higher k-effective level that would allow keeping the originally designed fuel composition and would introduce excess reactivity to compensate for fuel burn-up. A trend that could be observed in the advanced 3D model developed this year was in fact a lower keffective level for reactor nominal conditions that should be compensated, were this fuel composition to be used in the SCFR, with a higher enrichment or a different core design. TABLE VII. Void effect estimation: advanced heterogeneous model (tight blankets). gap=1 mm
k-eff
+/-
[pcm]
[$]
nominal
1.02218
56
0
full void
1.00369
48
-1802
-4.9
gap=2 mm
k-eff
+/-
[pcm]
[$]
nominal
1.02350
59
0
full void
1.00134
54
-2162
-5.8
VI. SIMMER-III/MCNP4C COUPLING
Fig. 6. Advanced heterogeneous model (central clusters). TABLE VI. Void effect estimation for the advanced heterogeneous model. [$]
gap=0 mm
k-eff
+/-
[pcm]
nominal
1.02553
57
0
full void
1.02072
51
-460
-1.2
gap=1 mm
k-eff
+/-
[pcm]
[$]
nominal
1.02530
47
0
full void
1.01883
55
-619
-1.7
gap=2 mm
k-eff
+/-
[pcm]
[$]
nominal
1.02356
56
0
full void
1.01813
53
-521
Global 2003
New Orleans, LA
The results outlined so far showed the importance of the introduction of coolant density axial profiles (the water density inlet/outlet ratio is about 9), and hence of a detailed spatial distribution of the physical properties, on the accuracy of the estimation of the reactor safety parameters (e.g.: for ENDFB-VI.7 a temperature variation of +900K causes a void effect variation of ~+1$). It was thus decided to couple MCNP4C [7] with a thermalhydraulics code (SIMMER-III [8]) that could provide steady state coolant density and fuel temperature profiles, taking into account previously validated neutronics feedback [9]. The full coupling of the two system codes has not been implemented yet, as it would slow down excessively the overall computational time. A SIMMER-III steady state problem was run instead and then the calculated temperature and density maps were introduced in the MCNP4C model.
-1.4
November 16-20, 2003
1759
1 0.9 0.8 w er rela tive po
0.7 0.6 0.5 0.4 0.3
S25
0.2
S21
0.1
S17
0
Downcomer/RR (7)
ZrH 4-1 (35)
Blanket 4-2 (36)
Blanket 4-1 (34)
Seed 3-2 (31)
Seed 3-3 (32)
Lower Blanket (3) Lower Reflector (2) Inlet (1)
Fig. 7. SIMMER-III R/Z model. The coolant temperature profile was introduced making use of the MCNP ‘tmp’ card, which affects the scattering laws, but not the thermal matrices. Thermal scattering was considered at the average temperature of 800K, and so was the coolant in the blankets. No axial profile was introduced in the blanket regions, since only little power is produced in these regions (see Fig. 8) and the temperature profile is expected to be rather smooth. The application of SIMMER-III to the SCFR required the introduction of the equations of state for supercritical water and of its thermophysical properties. The validation of these new code features has not yet been New Orleans, LA
25
S1 28
19
22
Seed 3-4 (33)
mes h
Fig. 8. SIMMER-III power distribution
Seed 3-1 (30)
ZrH 3-2 (28)
Blanket 3-3 (29)
ZrH 3-1 (26)
Blanket 3-2 (27)
Blanket 3-1 (25)
Seed 2-3 (23) Seed 2-2 (22) Seed 2-1 (21)
ZrH 2-2 (19)
Blanket 2-3 (20)
ZrH 2-1 (17)
Blanket 2-2 (18)
Blanket 2-1 (16)
Seed 2-4 (24)
Seed 1-4 (15) Seed 1-3 (14) Seed 1-2 (13) Seed 1-1 (12)
ZrH 1-1 (10)
Blanket 1-2 (11)
Blanket 1-b (9)
Blanket 1-a (8)
13
radia l
Upper Blanket (4)
16
7 10
Outlet (6) Upper Reflector (5)
Global 2003
S9 S5
ax ial me sh
S13
1
SIMMER-III was developed by Japan Nuclear Cycle Development Institute (JNC) in collaboration with Forschungszentrum Karlsruhe (FZK), Commissariat à l'Energie Atomique (CEA) and Institut de Radioprotection et de Sûreté Nucléaire (IRSN). SIMMER-III is a twodimensional, three-velocity-fields, multi-phase, multicomponent, Eulerian fluid dynamics code coupled with a structure model (fuel pins etc.) and space-, time- and energy-dependent neutron dynamic model. SIMMER-III uses an elaborate scheme of equations of state functions for fuel, steel, coolant (light and heavy liquid metals, water and gas), absorber and simulation materials [10]. In neutronics, the transient neutron flux distribution is calculated with the improved quasi-static method [11]. For the space dependent part, a TWODANT-based flux shape calculation scheme has been implemented [12]. The SIMMER-III SCFR nodalization model is shown in Fig. 7. 28 radial and 26 axial fluid-dynamics meshes were used, while two different neutronics meshing grids were applied: 56x56 and 84x84. The active core is divided into 20 axial meshes; consequently a new MCNP model with 20 separately defined axial cells for each seed area was developed. 12 different materials were defined for the 12 different seed regions and, therefore, 12 different cross section sets are processed with an NJOY [13] batch routine, one for each region average temperature, and applied to the calculations.
completed and furthermore no final consensus has been reached on the general trends of supercritical heat transfer [14]. Therefore the results that will be hereafter presented have to be considered as preliminary and will need further validation, nonetheless it is believed that they show significant trends and demonstrate the importance of this coupled model and the specific relevance for its application to very heterogeneous reactors like the SCFR. A SIMMER-III calculation was then performed and the calculated temperature and density maps were subsequently introduced in the MCNP model. The results of the void reactivity effect calculations are illustrated in Table VIII.
4
VI.A. SIMMER-III/MCNP4C Model
November 16-20, 2003
Besides the void effect coefficient itself, which is now strongly negative beyond the uncertainties mentioned in [3], it is interesting to note the nominal conditions keffective, close to 1 and to the value calculated by TWODANT (k-eff = 1.01515522). The previously calculated k-effectives were in the order of ~0.97, which, were they accurate, would have implied a review of the core design and/or of the fuel composition, which already consists of an average enrichment of ~23% in plutonium. TABLE VIII. Void effect: coupled calculation. gap=2 mm
k-eff
+/-
[pcm]
nominal
1.00277
66
0
full void
0.97308
47
-3043
[$] -8.2
The importance of including fuel and coolant temperature and density distributions in nuclear reactor core calculations is a known fact. The results reported in Table VIII show that the implementation of a fluid1760
dynamics/neutronics coupled system that can calculate accurately temperature and density distributions together with reactor k-effective is an important step for the prediction of modern reactors nominal conditions and static reactivity coefficients. VII. CONCLUSIONS The SCFR is a modern reactor concept with transmutation capabilities that because of its specific characteristics: supercritical fluid as a coolant, high power density, once-through cycle, heterogeneous core design and the potential for a conversion ratio greater than 1, requires the development of new tools and to some extent of a new approach for the analysis of its design and the evaluation of its safety related features. In this context a detailed MCNP model was developed and applied for the evaluation of the reactor void coefficient. The void effect calculations outlined large discrepancies among the different adopted models and nuclear data libraries. It is, however, not too surprising that the results do not entirely agree. It was argued in [3 and 9] that the void reactivity coefficient could be estimated to be approximately equal to ~0±2.5$. However considering that the average uncertainty associated with those MCNP calculations was of about 75pcm (~0.25$), and that the uncertainties related to the cross section estimations, evaluations, and measurements can be as high as 100%, it was possible to conclude that the uncertainty range was of around ±5$. Therefore unless more refined data and models were applied, confirmed by associated experimental evidence e.g.: by measurement in zero power critical facilities, and made available, in order to guarantee a negative void reactivity coefficient for the SCFR, the calculated value for this parameter should be of at least of –5$. Such a value would take into account the current uncertainties associated with the different cross section sets, the model and the code approximations, and guarantee with adequate confidence its negativity. With the intention then of improving the void reactivity effect and reducing the calculation uncertainties several sensitivity studies were performed to investigate the influence of different design and modeling details such as: coolant density axial profiles, subassembly gap spacing, new cross section data, new blankets and new core configurations. These analyses showed the relevance of a correct estimation of the reactor nominal conditions and of the accuracy of the MCNP model. Hence they support the decision to apply SIMMER-III and to introduce new equations of state and thermophysical properties for supercritical water. The implementation of a tool that can calculate the reactor initial conditions together with a sophisticated Monte-Carlo geometrical model is believed to be a remarkable progress and an important step for the
Global 2003
New Orleans, LA
November 16-20, 2003
accurate and reliable prediction of modern reactors reactivity coefficients. The subsequent adoption of the described line of approach made it, therefore, possible to decrease the calculation uncertainties and to improve the performance of the SCFR under voiding transients, therefore reaching the conservative value of –5$ indicated as a safety prerequisite in [3]. Furthermore, the achievement of a sound steady state would make it possible in the future to perform detailed burn-up calculations and carefully assess the transmutation potential of this reactor concept. ACKNOWLEDGMENTS We would like to thank Dr. Andrei Rineiski, Prof. Thomas Schulenberg, and Mr. Frank Kretzschmar for their guidance and support, JAPCO for having contributed to the funding of this work (contract No. XI-0118). REFERENCES 1.
2.
3.
4.
5.
6.
7. 8.
Y. OKA, S. KOSHIZUKA, Design concept of OnceThrough Cycle Supercritical-Pressure Light Water Cooled Reactors, The First International Symposium on Supercritical Water-cooled Reactors, Design and Technology [SRC-2000], November 6-9, Tokyo, Japan (2000). T. MUKOHARA, S. KOSHIZUKA Y. OKA, Core Design of a High-Temperature Fast Reactor Cooled by Supercritical Light Water; Annals of Nuclear Energy, 26, 16, 1423 (1999). MAGNUS MORI, V. SINITSA, A. RINEISKI, Uncertainties in Void Effect Estimations for a SuperCritical Water Fast Reactor with Transmutation Capabilities, Proceedings of the Annual Meeting on Nuclear Technology [JK2003], p. 29, May 20-22, Berlin, Germany, Inforum GmbH (2003). Y. OKA, T. JEVREMOVIC, Negative Coolant Void Reactivity in Large Fast Breeder Reactors with Hydrogenous Moderator Layer, Annals of Nuclear Energy, 23, 14, 1105 (1996). S. TANAKA, Y. SHIRAI, M. MORI, K. YAMADA, Y. KATAOKA, Y. KOMANO, Plant Concept of Supercritical Pressure Light Water Reactor, Proceedings of ICONE5, 2346, May 26-30, Nice, France (1997). T. JEVREMOVIC, Y. OKA, S. KOSHIZUKA, Core Design of a Direct-Cycle Supercritical-WaterCooled- Fast Breeder Reactor, Nuclear Technology, 108, 24, (October 1994). J. F. BRIESMEISTER, MCNPTM – A General Monte Carlo N-Particle Transport Code, Version 4C, LA-13709-M (March 2000). SA. KONDO, H. YAMANO, T. SUZUKI, Y. TOBITA, S. FUJITA, X. CAO, K. KAMIYAMA, K. MORITA, E. A. FISCHER, D. J. BREAR, N. 1761
9.
10.
11. 12.
13. 14.
SHIRAKAWA, M. MIZUNO, S. HOSONO, T. KONDO, W. MASCHEK, E. KIEFHABER, G. BUCKEL, A. RINEISKI, M. FLAD, P. COSTE, S. PIGNY, J. LOUVET, T. CADIOU, SIMMER-III, A Computer Program for LMFR Core Disruptive Accident Analysis, JNC TN9400 2000-002 (2000). MAGNUS MORI, A. RINEISKI, V. SINITSA, W. MASCHEK, Monte-Carlo and Deterministic Models for Void Effect Calculations in the SuperCritical Water Fast Reactor, to be published in the Proceedings of GENES4/ANP 2003, September 1519, Kyoto, Japan (2003). Y. TOBITA, SA. KONDO, H. YAMANO, S. FUJITA, K. MORITA, W. MASCHEK, J. LOUVET, P. COSTE, S. PIGNY, Current Status and Application of SIMMER-III, an Advanced Computer Program for LMFR Safety Analysis, Proceedings of NTHAS2: Second Japan-Korea Symposium on Nuclear Thermal Hydraulics and Safety, October 1518, Fukuoka, Japan (2000). K. O. Ott and R. J. Neuhold, Nuclear Reactor Dynamics, ANS, La Grange Park, USA (1986). G. Buckel, E. Hesselschwerdt, E. Kiefhaber, S. Kleinheins and W. Maschek, “A New SIMMER-III Version with improved Neutronics Solution Algorithms,” FZKA 6290, Karlsruhe, Germany (1999). R. E. MacFARLANE, D. W. MUIR, The NJOY Nuclear Data Processing System, Version 91, LA12740-M, Los Alamos National Laboratory (1994). I. L. PIORO, H. F. KHARTABIL, R. B. DUFFEY, Heat Transfer at Supercritical Pressures (Survey), Proceedings of ICONE11, 36454, April 20-23, Tokyo, Japan (2003).
Global 2003
New Orleans, LA
November 16-20, 2003
1762
