Basic dynamics of graphite moderated LEU fueled MSRs

Transcription

1 UTK seminar, July 18th 2014 Basic dynamics of graphite moderated LEU fueled MSRs Dr. Ondřej Chvála Seminar overview Historical context and lessons MSR salt & lattice choices Reactor dynamics: dρ/dt Focus on the graphite moderator NB: Thanks to Terrestrial Energy Inc. 1

2 ORNL Aircraft Nuclear Reactor Progress ( ) 1949 Nuclear Aircraft Concept formulated 1954 Aircraft Reactor Experiment (ARE) built and operated successfully (2500 kwt2, 1150K) 1951 R.C. Briant proposed LiquidFluoride Reactor 1952, 1953 Early designs for aircraft fluoride reactor MWt Aircraft Reactor Test (ART, Fireball ) proposed for aircraft reactor 1960 Nuclear Aircraft Program canceled in favor of ICBMs 2

3 The Aircraft Reactor Experiment (1954) In order to test the liquid-fluoride reactor concept, a non-circulating core, sodium-cooled reactor was hastily converted into a proof-of-concept liquid-fluoride reactor. The Aircraft Reactor Experiment ran for 1000 hours at the highest temperatures ever achieved by a nuclear reactor (1150 K). Operated from 10/30/1954 to 11/12/1954 Liquid-fluoride salt circulated through beryllium reflector in Inconel tubes. 235UF dissolved in NaF-ZrF 4 4 Produced 2.5 MW of thermal power. Gaseous fission products were removed naturally through pumping action. Very stable operation due to a large negative temperature-reactivity coefficient. Demonstrated load-following operation without control rods. 3

4 Aircraft Nuclear Program Allowed ORNL to Develop Reactors It wasn t that I had suddenly become converted to a belief in nuclear airplanes. It was rather that this was the only avenue open to ORNL for continuing in reactor development. That the purpose was unattainable, if not foolish, was not so important: A high-temperature reactor could be useful for other purposes even if it never propelled an airplane Alvin Weinberg 4

5 Molten Salt Reactor Experiment ( ) ORNLs' MSRE: 8 MW(th) Designed Started in 1965, 5 years of successful operation Developed and demonstrated on-line refueling, fluorination to remove uranium UF4+F2 UF6, Vacuum distillation to clean the salt Operated on all 3 fissile fuels U233, U235, Pu239 Some issues with Hastelloy-N found and solved Further designs suggested (MSBE, MSBR, DMRS), none built After Alvin Weinberg was removed from ORNL directorate, very little work done, almost no funding The Molten Salt Reactor Adventure, H. G. MacPherson, Nuclear Science and Engineering 90, p (1985) 5

6 Current Predicament ORNL's program in the 1960s was predicated on many historical circumstances, which are not valid any more. Current political priorities: inherent walkaway safety, proliferation resistance, TRU actinide minimization and spent nuclear fuel inventory management, among others. Economic necessity: minimization of upfront costs, maximization of resource utilization (see later), and exploring new markets. Any futuristic R&D program needs to get actually funded. Any new reactor R&D and deployment (R&D&D) needs: to get regulated using the standard rules tailored to LWR significant but not insurmountable challenge, necessitates new generation of experts in related areas. 6

7 Salt Selection Salts composition differences impacts salt cost, tritium production, and neutronic performance. THD differences neglected. 7

8 Salt/Graphite Lattice Parametric scan of neutronics with different salts. Infinite hexagonal lattice of graphite and salt fuel. Lattice parameters are channel pitch p and salt fraction f. 3 f Salt channel radius: r = p 2 2 2π Reflective unit cell: FUEL SALT GRAPHITE MODERATOR 8

9 Temperature-Reactivity Feedbacks Reactor dynamics is crucial to safety. More absorptive salts are problematic in FHRs. MSRs? k 1 Reactivity: ρ = k dρ Temperature-reactivity coefficient =? dt needs to be negative. Three components to the thermal feedback in MSR: Doppler due to resonance broadening: prompt feedback Salt expansion with temperature: fast feedback Whole core expansion: slow feedback Must investigate finite cores, as changes in leakage may be significant. 9

10 Finite Core Geometry Our play reactor is a graphite block with holes, surrounded by 5cm down-comer and 15cm top&bottom plenum, inside a 3cm thick Hastelloy-N tank. Graphite block sizes studied: 2x2, 3x3, 5x5, 7x7 meters. Channel pitch and radius from the lattice searches. Graphite Salt Tank Top view Side view 10

11 Calculation Strategy Using MCNP5 MCNP5 does not broaden the cross-sections on the fly. NJOY used to create a special Doppler-broadened crosssection library for every 50 K step. Three cases investigated for every salt selection: a) Doppler effect only just use a different cross-section library for the fuel salt depending on the temperature. b) Doppler + salt expansion the above, plus salt density decreased by 10-3 g/cm3 per K c) Doppler + salt expansion + thermal expansion of the whole reactor the above, plus physical expansion and temperature changes of the graphite moderator and the tank. Graphite linear thermal expansion coefficient = 3.5E-6 m/m per K Hastelloy-N linear thermal expansion coefficient = 15.0E-6 m/m per K 11

12 Calculation Strategy Using MCNP5 II For every salt, two lattice geometries (i.e. the channel pitch p and salt fraction f) are investigated: 1) p and f at max FoM from infinite lattice studies above; 2) f at max FoM given p=15cm using the studies above. First, find critical uranium enrichment level of the finite core at 900K for every salt and core geometry. Then, for each case, re-calculate keff at temperatures T=800K, 850K, 950K, 1000K, and 1100K. k 1 ρ = k From keffs, calculate reactivity at every T: Fit the reactivity slope as fcn(t) for every case: dρ ρ(t ) = T + c dt 12

13 72% 7LiF 16% BeF2 12% UF4 13

14 73% 7LiF 27% UF4 14

15 78% NaF 22% UF4 15

16 49% NaF 38% ZrF4 13% UF4 16

17 58% NaF 30% BeF2 12% UF4 17

18 74% NaF 12% BeF2 14% UF4 18

19 46% NaF 33% RbF 21% UF4 19

20 50.5% NaF 21.5% KF 28% UF4 20

21 Summary: dρ/dt as a function of the core size All temperature-reactivity feedbacks are negative for all the cores investigated. Doppler feedback depends on the salt choice, independent of the core size. Salt thermal expansion enhances feedbacks for small cores (for 3m enhances or is about the same), it reduce feedback for larger cores. The effects of whole core (i.e. graphite) thermal expansion adds little to salt expansion feedback, in most cases insignificant enhancement. This is surprising since the French group found a large effect from graphite expansion let us investigate. 21

22 Isolating the graphite contribution to dρ/dt What happens when the core gets hotter? The salt will be kept at 1100K in all calculations. The graphite and tank will increase in temperature (including dimension and density effects) from 800K to 1100K. Imagine that the salt heats up by a reactivity insertion, and the rest of the core starts to warm up consequently. Then for every core graphite temperature we calculate reactivity for all salt the selections, core sizes from 2m to 7m, and two different lattice configurations (FoM and FoM15cm). 22

23 Calculation Strategy Using MCNP5 III Since graphite serves as a moderator, thermal scattering S(α,β) has to be taken into account. In prior work, the S(α,β) libraries used were all at 1000K (grph.16t). Unfortunately we do not have 50K interpolation for thermal scattering libraries as we have for continuous energy cross-sections (work in progress). For this update, S(α,β) libraries shipped with MCNP5, evaluated at 800K, 1000K, and 1200K were used: Graphite temperature [K] Temperature used for S(α,β) [K] MCNP5 S(α,β) library grph.15t grph.15t grph.16t grph.16t grph.16t grph.17t 23

24 FoM 2m x 2m cores FoM p=15cm Salt-averaged dρ/dt from K 24

25 FoM 3m x 3m cores FoM p=15cm Salt-averaged dρ/dt from K 25

26 FoM 5m x 5m cores FoM p=15cm Salt-averaged dρ/dt from K 26

27 FoM 7m x 7m cores FoM p=15cm Salt-averaged dρ/dt from K 27

28 Conclusions Strong negative temperature-reactivity feedback of graphite moderator seen for core sizes from 2m to 7m & all salt selections, caused by thermal scattering temperature dependence. The smaller the core the stronger the effect (leakage). This contradicts previous research. However it is to be expected that a hot moderator is less efficient than a cold one, thus reducing moderation with increased temperature. The slides have been posted: Thank you for your attention! 29