Two Methods for Converting Iran s IR-40 Reactor to Use Low-Enriched-Uranium Fuel to Improve Proliferation Resistance After Startup R.S. Kemp March 2014 Abstract This article demonstrates the feasibility of converting Iran s 40MWt heavy-water research reactor at Arak (the IR-40) to use low-enriched uranium fuel in order to slow the production of weapon-usable plutonium. Two methods are proposed, and both retain identical power density, thermal-hydraulic, and safety profiles as the original design and require no physical alteration of the core. They can be implemented at any time during the reactor s life. The two methods have competing effects on cycle length, and can be combined to produce an optimized core that matches the start-of-cycle reactivity and cycle length of the original reactor. The methods are flexible enough to meet politically imposed demands, such as the need to consume a pre-specified amount of domestically produced enriched uranium. The proposed optimal design produces weapon-grade plutonium at only about 19 percent of the rate of the unmodified reactor for the same power level, cannot be readily converted back to natural-uranium fuel, and retains the ability to produce medical isotopes at rates that exceed the original design through the use of LEU targets. 1 Policy Context for Conversion Iran can provide confidence that its nuclear program is peaceful by adjusting its nuclear infrastructure so the time needed to produce a weapon quantity of nuclear-explosive fissile material is longer than some minimum, diplomatically negotiated period. In this respect, Iran s program contains two areas of primary concern: a capability to produce high-enriched uranium using centrifuges installed at Natanz and Fordow, and an emerging plutonium capability associated with nuclear reactors, and in particular the IR-40 heavy-water reactor under construction at Arak. Department of Nuclear Science and Engineering, Massachusetts Institute of Technology Address correspondence to rsk@mit.edu. The author thanks Kord Smith for assistance in modeling the IR-40 reactor with CASMO. 1
The original design of the IR-40 reactor uses natural (non-enriched) uranium and heavy water as a moderator. Reactors of this design produce plutonium quickly and have been used to support nuclear-weapon programs in the past. If the IR-40 were made to use LEU fuel instead, the reactor could be made to produce weapon-grade plutonium more slowly (at about 19 percent of the rate for the design proposed here, or less if usage demands allow the power to be reduced) providing a significant nonproliferation advantage in terms of Iran s plutonium capability. The uranium and plutonium aspects of Iran s program, while nominally distinct, may become linked if Iran s enrichment program is to support Iran s nuclear reactors. Iran has two operational reactors requiring regular supplies of low-enriched uranium (LEU): the Bushehr light-water power reactor and the Tehran Research Reactor (TRR). The Bushehr reactor is already supplied with fuel under a ten-year contract, and uses a proprietary Russian-made VVER fuel that Iran cannot credibly make at this time. The Tehran Research Reactor uses 19.75-percent enriched fuel, but at this time Iran has already produced enough 19.75-percent enriched uranium to run the TRR for an additional ten years at nearcontinuous operation. Thus neither of Iran s two reactors designed to consume enriched uranium can do so in the near term, and thus neither provides a basis for sizing Iran s enrichment program presently under negotiation. Iran has plans for additional LEU-fueled reactors, but those reactors are not yet designed. Given typical timelines, they are not likely to be realized for at least a decade. If no near-term destination for Iranian LEU is identified, and unused stocks of LEU were allowed to accumulate, this would significantly reduce the time needed to make a weapon by the enrichment route. 1 To prevent this, the accumulating LEU could be exported, or alternatively, the modified IR-40 reactor could consume the LEU on an ongoing basis, becoming the only credible way Iran could domestically utilize the LEU produced by its enrichment program in the near term. A converted IR-40 reactor could, therefore, provide a rational basis for sizing the enrichment program as well. 2 Previous Studies The conversion of the IR-40 reactor has been previously studied by Thomas Mo Willig et al. The authors suggest increasing the enrichment and power density of the core, but maintaining identical reactor power at 40 MWt, by reducing the number of active fuel tubes from 150 to 60. 2 This modification increases the power density in the fuel region by 1 For example, a tiny stockpile of the LEU sufficient to fuel the Bushehr reactor for just one week would also reduce the time needed to make a nuclear weapon by about 3 months, assuming an enrichment capacity of 10,000 operating IR-1 centrifuges. See: http://lewis.armscontrolwonk.com/archive/6874/kemp-settinga-goal-for-iran-talks 2 Thomas Mo Willig, Cecilia Futsaether, and Halvor Kippe, Converting the Iranian Heavy Water Reactor IR-40 to a More Proliferation-resistant Reactor, Science & Global Security 20:97 116 (2012); DOI: 10.1080/08929882.2012.713767. 2
a factor of 2.5. To maintain an unchanged temperature profile in the fuel, the mass-flow rate of the coolant must increase proportionally. If the fuel geometry is also unchanged, the increased cooling requirement is achieved by increasing the coolant flow velocity by the same factor of 2.5. The pressure drop, under normal turbulent flow, is proportional to the velocity squared. The result is that the cooling system for the Willig modification would require cooling pumps with more than 6 times the pumping power planned for the original IR-40 design. This might also necessitate larger pipework, heavier control systems, different pressurizes, different safety strategies, etc. An alternative approach would be to use the Willig core design, or a similar such design, but reduce the power to no more than 16 MWt. This approach would preserve the cooling requirement of the reactor, and allow Iran to benefit from a higher peak neutron flux at the center of the reactor afforded by the more compact core design. In many ways, this approach might be regarded as the best conversion strategy, but as it would require modifications to physical infrastructure of the core (specifically, the plugging of unused fuel tubes and possible introduction of new channels), such a change would be most viable if implemented before the reactor became critical. Once the reactor operates and core infrastructure becomes radioactive, such modifications are much more difficult to implement. A study of options along these lines has been undertaken in parallel to this work. 3 If, for political reasons, one of the above methods of conversion cannot be implemented, then there are still fallback options that can be implemented easily at any time. This article examines two such options and their benefits. 3 Constraints on the Conversion of the IR-40 Reactor The conversion of the IR-40 reactor is subject to both political and technical constraints. Politically, conversion is more likely to be accepted by Iran if it requires few design changes and maximally uses existing investments. Iran s acceptance of conversion is also likely to be contingent on other factors in negotiations, such as an early demonstration of good faith by other negotiating parties. The timing of political constraints is nearly impossible to control and might not fortuitously align with the construction schedule of the IR-40 reactor. If the reactor becomes critical and thus radioactive, changes to the reactor s infrastructure will be more difficult, but conversion by the methods proposed here would remain viable and easy to implement. The international community has made clear that it prefers Iran to keep its domestic enrichment levels to less than 5 percent U-235. For this reason, this study will assume that the LEU fuel to be used in a modified IR-40 reactor will be enriched to 5 percent. However, it is worth noting that this might not be the optimal level. There is, in principle, a balance 3 Ali Ahmad, Frank von Hippel, Alexander Glaser, and Zia Mian, A Win-Win Solution for Iran s Arak Reactor? Arms Control Today, April 2014. 3
to be struck between the proliferation potential of the plutonium and uranium aspects of Iran s program, and the optimal enrichment level may be something else. However, given the difficulty of trying to optimize technical constraints through diplomatic talks, 5 percent is a reasonable and workable presumption. The primary technical constraints is that the conversion process must result in a safely operating reactor. The new reactor core must have sufficient reactivity to operate, but not so much that it cannot be safely controlled. In this study, we match start-of-cycle reactivity because we know that the existing reactivity-control mechanisms will then be sufficient. One could additionally consider scenarios with additional reactivity, which may be tolerated with the addition of burnable poisons and result in longer core life, but these are beyond the scope of this proof-of-concept study. A second neutronics constraint is that the reactor must be capable of fulfilling whatever scientific requirements it has been designed to execute. For example, the reactor may need to produce medical isotopes at a certain rate, and that will necessitate some minimum neutron flux in the channels in which isotope-producing targets are to be irradiated. One of the consequences of converting from natural to low-enriched uranium fuel is, for the same geometric buckling, the neutron flux will go down for any given power density. However, this can be compensated for by raising the enrichment of the isotope-producing targets. If the original reactor was designed for natural-uranium targets, then the use of LEU targets can compensate for the lower neutron flux, leaving medical-isotope production rates unchanged or improved. Third, the reactor is subject to thermal-hydraulic constraints. These include the performance of the primary-loop cooling system during full-power operation, and the safety systems designed to remove decay heat in the event the cooling system fails. Changes to the power distribution within the core can have significant impacts on these systems. In this paper, we consider only conversion options that preserve or reduce the cooling requirements of the reactor. Furthermore, we make no changes to the geometry, so as not to disrupt any passive cooling mechanisms that might exist. Finally, and least important, the reactor should have a practical cycle length. Under normal design constraints, a long cycle length that makes efficient use of the uranium fuel is preferred. However, in the case of the IR-40 reactor, it may be more important to optimize for nonproliferation goals rather than cycle length. In this study, we should that it should be possible to leave the cycle length unchanged from the original design. 4 Methods of Conversion Considered The purpose of conversion is to reduce plutonium production. At minimum, this requires moving from natural-uranium fuel to low-enriched fuel. There are two basic methods for doing this while leaving the core geometry and thermal hydraulics unchanged. 4
Figure 1: Photograph of tank of the IR-40 reactor showing fixed pressure tubes and dimensions consistent with data collected by Willig. Photo credit Yasaman Hashemi, PressTV, Iran. 5
1. Reduce the density of fissile isotopes in the core to approximately what would have been present if the reactor were fueled with natural uranium. Such a reduction can be effected by using a dispersion fuel, in which uranium dioxide is dispersed within a neutronically inactive filler, such as pure aluminum. Dispersion fuels are commonplace in research reactors. For example, the Tehran Research Reactor uses a dispersion fuel. The method of changing the enrichment level for a reactor by changing the fissile-material density is also common, and has been the mainstay of the RERTR (Reduced Enrichment for Research and Test Reactors) program, which has converted over 40 reactors to use LEU fuel worldwide. 4 2. Absorb neutrons to control excess reactivity arising from the increased enrichment. Because LEU introduces extra fissile material relative to natural uranium, the extra reactivity can be absorbed by neutron poisons in order to adjust the start-of-cycle reactivity to the maximum allowed value. Traditionally this is done by adding burnable poisons such as boron, gadolinium, or erbium. In the case of the IR-40 reactor, it is also possible to use an unusual but effective non-burnable poison: a small amount of chemically identical (and therefore perfectly compatible) light water added to the heavy water used as coolant and moderator. Light water acts as a poison because the protium-hydrogen absorbs reactivity. Unlike burnable poisons, light water will not extend the cycle length, but it makes it difficult to convert the reactor back to natural-uranium fuel without a time-consuming and highly visible process of flushing the moderator and replacing it with a fresh batch of high-purity heavy water. If heavy-water production is limited and monitored, the absence of pure heavy water in the reactor would ease safeguards and provide a meaningful technical delay to proliferation. Some combination of burnable and light-water poisons might provide an optimal balance between reversibility and practical aspects of cycle length. Both of the above methods will enable the reactor to use LEU fuel. They can be implemented at any time during the reactor s life, without any changes to hardware, and for very low cost. The remainder of this paper examines the viability and consequences of implementing these concepts. 5 Design of the IR-40 Reactor According to data compiled by Willig and an Iranian reactor-safety analysis, 5 the IR-40 is a tank-type reactor with pressure tubes, using heavy-water coolant and moderator, and 4 Normally the RERTR program is reducing enrichment from HEU to LEU by increasing the density; in the case of IR-40 reactor, the conversion goes the other way. 5 F. Faghihi, E. Ramezani, F. Yousefpour, S.M. Mirvakili, Level-1 probability safety assessment of the Iranian heavy water reactor using SAPPHIRE software, Reliability Engineering and System Safety 93:1377 1409 (2008); DOI: 10.1016/j.ress.2007.10.002. 6
150 RBMK-1000-style 19-pin fuel bundles in a hexagonal lattice with a bundle-to-bundle pitch of 26.5 cm. The reactor s design parameters are given in Table 1. In addition to these provided data, this analysis assumes that the pressure tubes and major structural elements of the RBMK fuel bundles are made of Zircaloy-2. It is also assumed that the center tube, which forms a key structural element of the fuel bundle, has the same inside and outside dimensions as the fuel-cladding tubes and is filled with coolant. Table 1: Key Parameters of the IR-40 Reactor Power 40 MWt Coolant and moderator composition D 2 O (99.75%) Coolant and moderator inlet pressure 3.5 bar Coolant and moderator inlet pressure 3.5 bar Coolant average inlet temperature 323 K Coolant average exit temperature 343 K Pressure-tubes with fuel 150 Pressure-tube lattice pitch 26.5 cm Pressure-tube inner radius 4.0 cm Pressure-tube wall thickness 0.4 cm Fuel composition and density 10.4 g/cc-uo 2 Fuel outer radius 0.5740 cm Clad inner radius 0.5965 cm Clad outer radius 0.6815 cm Clad and assembly material Zircaloy Fueled rods per bundle 18 Empty (water filled) rods per bundle 1 (center) Active fuel length 340 cm Power density in fuel (calculated) 4.58 W/gU Temperature of fuel (estimated) 373 K 6 Method of Analysis To demonstrate the feasibility of the two conversion concepts, we want to find (1) the fissile material density in fuel diluted with aluminum powder, or (2) the amount of light water to be added to the heavy-water moderator, that will produce the same start-of-cycle reactivity (or k eff ) with 5-percent enriched fuel as would have been achieved in the unmodified design. For the purposes of this proof-of-concept analysis, this paper uses a simple but adequate approach to reactor analysis based on an infinite-lattice calculation and a subsequent onegroup bare-core diffusion calculation. We proceed as follows: 7
1. Construct the infinite lattice of the fuel, coolant, and moderator geometries specific to the original IR-40 design shown in table 1 and numerically solve for k and the migration area M 2 at the start of cycle, taking into account reactivity loss to equilibrium xenon poisoning. We do this numerically with a 70-energy-group neutrontransport code. 2. Use the above result and a one-group bare-core diffusion model to compute k eff for the original IR-40 design. The use of one-group diffusion theory is particularly valid here because heavy water is a weakly absorbing medium with a large mean free path. 3. Repeat steps 1 and 2 for 5-percent enriched fuel, adjusting either fissile-material density (method 1) or the H 2 O:D 2 O ratio in the moderator (method 2) until we find solutions that produces a k and migration area M 2 tuple that, when inserted into one-group diffusion theory, results the same start-of-cycle k eff as the original IR-40 design. 6.1 Infinite lattice computation of k and M 2 To compute k and M 2, it is necessary to model accurately the geometry of the lattice described in table 1, and to take into account effects such as parasitic neutron capture in fission products and resonance self-shielding effects which go beyond what is possible with one-group diffusion theory. We do this with CASMO-4E, a reactor analysis code developed for power-reactor designers by Studsvik Scandpower, a division of Studsvik AB, Nykoping, Sweden. CASMO version 4E is capable of modeling infinite cluster geometries, like the hexagonal lattice of the IR-40, by using a white boundary condition in which neutrons crossing the boundary are assumed to be isotropically instead of spectrally reflected. 6 The model used here is therefore a single bundle with associated coolant and moderator, as shown in figure 2, surrounded by a white boundary, which is why the model geometry is circular rather than hexagonal. CASMO breaks the geometry into a fine mesh, shown in figure 3 and solves the neutron-transport equations across the mesh for 70 energy groups using ENDFB-VI cross-section libraries under the assumption that the neutron source is spatially flat and isotropic inside each mesh element. CASMO also computes burn-up isotopics, tallies the corresponding plutonium production and isotope vector, and can model extensive core physics such a doppler broadening and thermal expansion in materials. The result of this model is that the IR-40 infinite lattice has a k = 1.128 and M 2 = 408.8 at the start of cycle after equilibrium xenon poisoning (determined at 0.1 MWd/Kg burnup). Detailed results at different burnups are shown in table 2. 6 D. Knott, M. Edenius, J. Peltonen, and M. Anttila, Results of Modeling Hexagonal and Circular Cluster Fuel Assembly Designs using CASMO-4, Proceedings of the American Nuclear Society Topical Meeting Advances in Nuclear Fuel Management II, Myrtle Beach, March 23 26, 1997, EPRI TR-107728- V1. 8
Figure 2: Actual neutron tracks computed for a single fuel bundle and associated moderator used in the CASMO-4E infinite lattice calculation. Colors relate to different materials. The 18-pin fuel bundle (purple), cladding (grey), and pressure tube (black) are clearly visible at the center of coolant and moderator (blue). Note the tracks are very dense so as to appear almost solid; the white speckles are the residual areas not covered by tracks. 9
Figure 3: The same tracks shown in Figure 2, but with the color of each track changing at the boundary between computational mesh cells. Within each mesh cell the neutron sources are assumed to be spatially flat and isotropic. 10
Table 2: CASMO output for the unmodified IR-40 infinite lattice. Burnup k M 2 U-235 Fissile Pu Total Pu MWd/Kg w% w% w% 0.000 1.14416 403.81 0.720 0.000 0.000 0.100 1.11327 397.48 0.708 0.007 0.007 0.500 1.11354 393.88 0.664 0.038 0.039 1.000 1.11194 390.57 0.612 0.072 0.076 2.000 1.10118 386.50 0.521 0.126 0.139 3.000 1.08447 383.93 0.443 0.167 0.192 4.000 1.06484 382.48 0.375 0.199 0.237 5.000 1.04377 381.65 0.316 0.223 0.278 6.000 1.02225 381.19 0.266 0.242 0.313 7.000 1.00102 380.91 0.222 0.257 0.345 8.000 0.98062 380.67 0.185 0.268 0.374 9.000 0.96146 380.41 0.153 0.277 0.401 10.000 0.94380 380.06 0.126 0.285 0.425 6.2 Calculation of k eff by one-group diffusion theory We use one-group diffusion theory to convert k of the infinite lattice to k eff for a finite core. On group diffusion theory can be thought as the steady-state balance between production, consumption, and leakage of neutrons for a fixed-size reactor of a homogenous material. The continuity equation for an arbitrary volume is: N(r, t)/ t = J(r, t) + S(r, t) s(r, t) (1) where N is the neutron density, and J is the neutron current representing the leakage from the volume, S is the sum of all internal sources, and s is the sum of all internal sinks. Fick s law gives the relationship between the net neutron current (vector flux) J and the scaler flux-density φ: J(r, t) = D(r) φ(r, t) (2) Where D is the diffusion constant. In the steady-state condition, N/ t = 0, so we can rewrite equation 1 as simply the balance of sources, sinks, and leakage: 0 = [D(r) φ(r)] + S(r) s(r) (3) The one-group approximation means we estimate sources S and sinks s arising from nuclear reactions in the volume from the flux density φ and from the macroscopic interaction cross-sections Σ, which have been weighted over all energies represented by φ and all 11
materials within the reactor. We write the macroscopic absorption cross-section Σ a and the macroscopic fission cross-section as Σ f. We further assume neutrons are produced only by fission. The sources and sinks are then: S(r) = 1 νσ f (r)φ(r) k eff (4) s(r) = Σ a (r)φ(r) (5) where ν is average number of neutrons (prompt and delayed) produced by a fission event, and k eff is the effective neutron multiplication rate between generations. Substituting S and s into equation 3, assuming core materials are spatially uniform, making use of the geometrical buckling B 2 = [ 2 φ(r)]/φ(r), canceling φ(r), and solving for k eff gives: k eff = νσ f DB 2 + Σ a (6) Dividing through by Σ a lets us re-write in terms of k = νσ f /Σ a and the migration area M 2 = D/Σ a, both of which we found numerically using CASMO in section 6.1. k eff = k 1 + M 2 B 2 (7) Equation 7 gives us k eff and allow us to compare different fuel and moderator compositions under the assumption that the geometrical buckling B 2 remains unchanged. 7 Results 7.1 Method 1: Dispersion of fissile material in an aluminum matrix Using the process described in section 6.1, we iteratively search for the amount of aluminum that, when added to 5-percent enriched uranium dioxide, produces the same k eff as the unmodified natural-uranium reactor. The solution occurs at 0.526 g/cc uranium (heavy metal at 5% U-235) dispersed in 2.31 g/cc aluminum. Detailed results are given in table 3. 7.2 Method 2: Dilution of heavy-water moderator with light water Using the process described in Section 6.1, we iteratively search for the the amount of light water that, when added to heavy water, produces the same k eff as the unmodified natural-uranium reactor. The solution occurs at 81.5 atom-percent heavy water. 7 Detailed results are given in table 4. 7 The addition of approximately 20% light water will reduce coolant viscosity and density by less than 2%, a change that is likely too small to have a detrimental effect on any cooling system or safety profile and leaves the thermal hydraulics of the reactor effectively unchanged. 12
Table 3: CASMO output for the 5% enriched aluminum-dispersion fuel infinite lattice. Burnup k M 2 U-235 Fissile Pu Total Pu MWd/Kg w% w% w% 0.000 1.24578 705.84 4.940 0.000 0.000 0.100 1.21050 692.57 4.928 0.000 0.000 0.500 1.20059 692.65 4.879 0.004 0.004 1.000 1.19264 694.16 4.818 0.009 0.010 2.000 1.18062 698.72 4.697 0.022 0.022 3.000 1.16958 703.86 4.577 0.034 0.035 4.000 1.15844 709.18 4.456 0.046 0.047 5.000 1.14708 714.66 4.337 0.057 0.059 6.000 1.13548 720.30 4.218 0.068 0.071 7.000 1.12359 726.12 4.099 0.079 0.083 8.000 1.11140 732.10 3.981 0.090 0.095 9.000 1.09890 738.30 3.864 0.100 0.106 10.000 1.08607 744.58 3.747 0.109 0.118 Table 4: CASMO output for the light-water diluted IR-40 infinite lattice. Burnup k M 2 U-235 Fissile Pu Total Pu MWd/Kg w% w% w% 0.000 1.04691 121.28 5.000 0.000 0.000 0.100 1.03333 120.99 4.988 0.002 0.002 0.500 1.02651 120.89 4.940 0.010 0.010 1.000 1.02204 120.88 4.881 0.019 0.019 2.000 1.01557 120.89 4.763 0.038 0.039 3.000 1.00959 120.93 4.645 0.056 0.057 4.000 1.00363 120.96 4.529 0.073 0.075 5.000 0.99761 121.01 4.413 0.089 0.093 6.000 0.99150 121.06 4.299 0.105 0.110 7.000 0.98528 121.11 4.185 0.120 0.127 8.000 0.97891 121.16 4.072 0.134 0.144 9.000 0.97241 121.22 3.960 0.148 0.160 10.000 0.96576 121.28 3.849 0.161 0.175 13
7.3 Optimization: Combined dispersion-dilution reactor The above results demonstrate that it is technically feasible to convert the IR-40 reactor to LEU fuel by either of the two methods proposed. It is notable that the use of an LEU dispersion fuel extends the cycle life relative to the unmodified reactor (for the same initial k eff ), whereas the light-water dilution method severely shortens cycle life but offers additional nonproliferation advantages of making conversion more difficult to reverse. The two methods can be combined to achieve the same initial reactivity and same cycle length as the original design, while producing less plutonium and achieving the extra nonproliferation advantages afforded by the use of light water. This occurs at approximately 95.8 atompercent heavy water and 1.01 g/cc uranium (heavy metal at 5% U-235) dispersed in 2.18 g/cc aluminum. Detailed results are shown in table 5. Table 5: CASMO output for reactivity and cycle-length matched IR-40 infinite lattice with extra proliferation resistance. Burnup k M 2 U-235 Fissile Pu Total Pu MWd/Kg WT% WT % WT % 0.000 1.11494 367.95 4.940 0.000 0.000 0.100 1.08617 363.26 4.928 0.000 0.000 0.500 1.07841 363.26 4.879 0.005 0.005 1.000 1.07177 363.74 4.819 0.012 0.012 2.000 1.06130 365.22 4.698 0.025 0.025 3.000 1.05127 366.88 4.577 0.038 0.039 4.000 1.04113 368.61 4.457 0.050 0.052 5.000 1.03080 370.41 4.338 0.062 0.064 6.000 1.02026 372.27 4.219 0.074 0.077 7.000 1.00948 374.18 4.101 0.085 0.089 8.000 0.99845 376.14 3.984 0.095 0.101 9.000 0.98712 378.17 3.867 0.106 0.113 10.000 0.97553 380.25 3.751 0.116 0.125 8 Non-Proliferation Consequences of Conversion Using the plutonium tallies in the CASMO outputs given in tables 2 5, we can estimate the plutonium production rates and plutonium-isotope vectors of the various core designs. Two immediate effects are observable: the plutonium production rates in the LEU cores are much lower, but the isotopic vectors in the LEU reactors retain a more weapon-favorable 14
fraction of fissile isotopes at higher burnups. This means there is a tension between overall plutonium production rates and the weapon usability of the plutonium. 8 Weapon-grade plutonium is defined as plutonium having less than 7 weight-percent Pu-240. 9 From the CASMO outputs, we fitted plutonium production rates and isotope vector trends to deduce the effective rate at which each reactor could produce weapongrade plutonium. The results show the reduced plutonium-production rate arising from the use of LEU fuel substantially outweighs the negative effect of the improved isotope vector. The maximum rates at which weapon-grade plutonium can be produced by each reactor configuration are shown in table 6. Table 6: Weapon-grade plutonium production in the IR-40 infinite lattice. Burnup Target Total Pu at WgPu Rate Relative Configuration for 7% Pu-240 Target Burnup at 40MWt Production [MWd/kg] [w%] [gpu/day] Rate Unmodified 1.55 0.103 26.64 1 Optimized 9.86 0.124 5.04 0.19 Moderator Dilution Only 8.64 0.155 7.20 0.27 Dispersion Fuel Only 10.3 0.121 4.72 0.18 Table 6 shows the optimized and dispersion-fuel cores have similar properties in terms of weapon-grade plutonium production potential, however the optimized core, being moderated by 95.8 weight-percent heavy water, will not sustain a chain reaction if refueled with natural-uranium fuel (k eff =0.95 for the infinite lattice without xenon). This small amount of light-water in the moderator prevents Iran from converting the reactor back to a more proliferation-capable reactor without first producing a fresh batch of moderator and coolant. Because heavy-water production is visible, slow, and easily monitored, this would not go unnoticed. 10 Better still, Iran s existing heavy-water production facility could 8 Modern weapon designs are not neutronically sensitive to plutonium isotopics, although heat from the decay of the higher plutonium isotopes may present a more practical, but not insurmountable, design challenge. The sophistication of a posited Iranian weapon design is not known to the author, but some evidence in the public domain suggests Iran may possess design knowledge that exceeds historical first-generation weapons. See: Nonproliferation and Arms Control Assessment of Weapons-Usable Fissile Material Storage and Excess Plutonium Disposition Alternatives, U.S. Department of Energy, DOE/NN-0007, Washington DC, January 1997, Box 3-1 (pp. 37 39); Mark Harrington, Evidence emerges of Iran s continued nuclear weapons research, Jane s International Defence Review, 13 March 2008. 9 U.S. Department of Energy, Plutonium: The First 50 Years, Report DOE/DP-0137, February 1996, http://fissilematerials.org/library/doe96.pdf. For our calculations, we assume CASMO s non-fissile plutonium tally (which is the sum of all even numbered isotopes) is effectively equivalent to a Pu-240 tally. The two will be very nearly equal at the low burnups considered here. 10 The Heavy Water Plant in Iran is rated at 16 tonnes/year. The Arak reactor needs about 100 tonnes of 15
be dismantled as a confidence-building measure with no negative impact on Iran s nuclear program (a small reserve of 95.8 weight-percent heavy water could be kept on hand for topping up the IR-40 reactor as needed). The combined dilution-dispersion strategy used for the cycle-length optimized core therefore provides a meaningful additional barrier to proliferation. Finally, it is important to note that although plutonium production has been slowed by roughly a factor of five, eventually the accumulated spent fuel will contain a weaponquantity of plutonium that could be used to make a weapon quickly. It will, therefore, be necessary to export spent fuel on an occasional basis. The frequency of export depends on the power and duty cycle at which the reactor is operated, but the modifications suggested here will significantly ease the export requirement. For example, if Iran used its 20-percent enriched uranium as targets for medical isotope production, the reactor could be operated at a continuous average power below 10MWt and still produce medical isotopes at a rate equal to or above that of the original IR-40 design with natural-uranium targets. Under this arrangement, the reactor would produce plutonium at less than 5 percent the rate of the original design and fuel would need to be exported no more often than once every 4 years (assuming a maximum allowed plutonium inventory in spent fuel of 4 kg). In practice, with additional isotope production ongoing at Iran s TRR reactor, the IR-40 reactor might operate at a lower duty cycle, and plutonium production and exports would be proportionally lower and less frequent. 9 Conclusions This article demonstrates the viability of converting the IR-40 reactor at Arak to use lowenriched uranium by two methods that can be effected after the reactor becomes critical, and is therefore not sensitive to the timing of the negotiations. By combining the two methods, it is possible to produce a 5-percent enriched LEU core with the same start-ofcycle reactivity and cycle length as the original, unmodified reactor, but which produces weapon-grade plutonium at only 19 percent the rate of the original reactor design. This optimized core has the additional advantage of preventing Iran from easily converting the reactor back to natural-uranium fuel. The foregoing analysis was constrained by the assumption that the LEU fuel had to be enriched to 5-percent U-235. Higher enrichments, if tolerable within the context of the enrichment program, would result in even less plutonium. Conversely, lower enrichments could be used to reduce the threat from the enrichment program at the cost of somewhat heavy water. International Atomic Energy Agency, Implementation of the NPT Safeguards Agreement and relevant provisions of Security Council resolutions in the Islamic Republic of Iran, 22 May 2013 (GOV/2013/27), paragraph 36; International Atomic Energy Agency, Implementation of the NPT Safeguards Agreement and relevant provisions of Security Council resolutions in the Islamic Republic of Iran, 28 August 2013 (GOV/2013/40), footnote 38. 16
greater plutonium production rates. The optimal configuration depends on the size of the enrichment program, and policies regarding the accumulated LEU it produces. This analysis did not consider other reactor of fuel-cycle optimization strategies. For example, cycle life could be extended by the use of burnable poisons and greater initial fissile loading. Neutron flux across the reactor could be flattened or tailored using different enrichments or U:Al dispersion ratios. To model such effects would require from Iran a clear definition of the reactor s purpose and performance requirements, as well as a finitereflected core model. These analyses are best performed jointly by Iran and its negotiating partners and beyond the scope of this analysis. Nevertheless, this paper demonstrates the feasibility and tremendous flexibility that exists in converting Iran s existing IR-40 reactor to a more peaceful configuration. 17