International Network of Engineers and Scientists Against Proliferation

Emerging Nuclear Technologies

The Example of Carlo Rubbia's Energy Amplifier¹

Christoph Pistner

Since the beginning of the 1990's, the concept of accelerator driven, subcritical nuclear reactors has gained renewed interest at different places all around the world. One major CERN group suggested a so called `Energy Amplifier' (EA), a cyclotron driven reactor, based on thorium fuel and a fast neutron spectrum.

This paper tries to give a short overview of this concept, as a general example for accelerator driven systems. Especially the main promises, like high proliferation resistance, high grade of inherent safety, no need for geologic storage and a possible elimination of nuclear waste will be reviewed in more detail.

The Promises

Currently, the development of nuclear power is stagnating, at least in the western world. This is due to various reasons, among them concerns about the final disposition of nuclear wastes, the risk of severe accidents and proliferation aspects of the fuel cycle as well as economical reasons.²

Nonetheless, new reactor concepts are developed to overcome some or all of these main obstacles of nuclear energy. Apart from `conventional' types of reactors like advanced liquid metal or high temperature reactors, new concepts like accelerator driven systems are investigated in more detail [Art 95], [Con97].

A group of scientists from CERN are proposing a so-called Energy Amplifier (see for example [Rub 94], [Rub 97a], [Rub 97b]) and are praising several advantages of this system:

"1. Extremely high level of inherent safety.

2. Minimal production of long lived waste and elimination of the need of geologic repositories.

3. High resistance to diversion, since latent proliferation is a major concern.

4. More efficient use of a widely available natural fuel, without the need of isotopic separation.

5. Lower cost of the heat produced and higher operating temperature than conventional PWRs...

Our design of an EA has these objectives as goals and it is intended as proof that they can be met fully". [Rub 97b, p. 188].

The aim of this paper is therefore to give an overview of the proposed concept with special regard to the main differences to other reactor systems and to discuss the question, wether the promised goals seem to be realistically achievable.

The Project

Any accelerator driven system consists of a subcritical core containing fissile material, which should not be able to sustain a chain reaction on its own. It will therefore need an external neutron source, produced by a beam of high energy protons (of the order of 1 GeV) hitting a target of heavy metal thereby producing a large number of neutrons by spallation reactions.

The main features of the Energy Amplifier concept shall be explained in brief. For more details, the reader is referred to [Rub 97b].³

The system is supposed to be driven by a cyclotron with a proton current of approximately 12.5 mA.⁴ The subcritical core will be made of thorium fuel, initially containing for example plutonium from spent LWR fuel as fissile material. The whole system is designed for an average value of the multiplication factor⁵ k_eff of 0.98; an average beam power of 30 MW will then produce a thermal power of 1500 MW.⁶

A fast neutron spectrum is chosen for the EA in order to maximize the fission probability of the actinide inventory and to minimize the neutron losses due to parasitic capture in fission products. Therefore, lead is chosen as a primary coolant and at the same time as spallation target for the proton beam (no separated spallation target).

The heat produced in the central core will be removed by natural convection of the primary coolant. Therefore the main vessel will be designed extremely tall in order to ensure a stable lead flow.⁷

No fuel re-shuffling or reloading is planed for the whole lifetime of the fuel, which is estimated to be 100—150 GWd/t (approximately 5 years).⁸ Afterwards the fuel will be reprocessed, the whole actinide inventory being reintroduced into the EA, only fission fragments are dedicated to final storage.⁹

Changes in the value of k_eff during burn-up, which will result in a change in the produced thermal power, are compensated by a variation of the proton current delivered to the system. Therefore no control rods are used in the EA.

Finally, three different passive systems to ensure inherent safety of the device are planned. They become active, if a temperature increase leads to a dilatation of the lead which will then fill special regions of the reactor through overflow paths. The first system is an emergency beam dump volume (EBDV) to stop the proton beam far away from the core in case of an accelerator shutdown failure. The Reactor Vessel Air Cooling System (RVACS), consisting of a narrow gap between the containment and the main vessel, which is normally filled with thermally insulating helium gas. If this gap is filled with lead, heat from the inner core is removed via natural convection of air to ensure the decay heat removal. At least a scram device of B₄C absorbers will be pushed into the core by overflowing liquid lead descending narrow tubes and thereby lifting up the B₄C rods.

The Problems

In this paragraph a short overview of open problems related to the EA shall be given, as can be found in the literature (a very comprehensive discussion is given for example in [HSK 97]. Of course, no complete list of unresolved questions can be given, nor can all of them be discussed in detail. Some important aspects will be investigated more deeply hereafter. The discussion is based on the concept as given in [Rub 97b].

• The reactor vessel must be fully accessible for in-service inspections. This is necessary for both the inner and outer vessel of the reactor. A relevant aspect is therefore that the coolant (lead) is not transparent and makes inspections more difficult, and that the double vessel hinders access to the inner part (RVACS).
  
  
• The major inherent safety systems (overflow path for expanding lead leading to the rise of B₄C-rods and the filling of the containment vessel RVACS) are neither testable, nor is a reset procedure given. This is especially evident for the RVACS. The reseting of the scram systems (at least in the case of B₄C-rods) will involve mechanical parts, which may lead to failure.
  
  
• The EBDV, which is positioned at the upper lead level near the main containment dome will act as a strong spallation neutron source in the vicinity of the containment dome.
  
  
• The control-rod free concept will lead to a highly heterogeneous burn-up. It is therefore questionable, whether fuel shuffling will become necessary or at least attractive for better fuel utilization.
  
  
• During the proposed burn-up of five years, large variations in k_eff are inevitable. Furthermore the reactivity "stored'' in the Protactinium-233, which will decay after reactor shut-down to Uranium 233 with a time constant of 27 d, is of the order of dk_eff=0.02. These effects strongly influence the necessary level of subcriticality. This problem will be discussed in more detail later on.
  
  
• Since the accelerator is used to maintain a constant power-level during burn-up (necessary because of the large variations in k_eff), it must be possible to vary the proton current in the range from approx. 5—20 mA. One is therefore led to the problem of ensuring that no uncontrolled increase in accelerator current is possible, while the system is in a state of high k_eff. This would lead to a sharp increase in reactor power therefore possibly leading to lead boiling and to core damage. This problem also arises evidently during any start-up of the reactor.
  
  
• Material problems due to the highly corrosive nature of lead still have to be solved. Especially the influence of impurities in the lead, produced by the spallation reactions in the central core, is not yet clear.
  
  
• The melting temperature of lead is 328° C. Therefore a freezing of lead during a reactor shutdown or cooldown transients has to be avoided. Freezing or at least partial freezing of lead can lead to damage of the primary heat exchangers or even a blockage of cooling channels in the reactor core, thereby resulting in core damage and loss of core geometry.
  
  
• Since the EA is operated with fast neutrons, a hypothetical loss of core geometry may lead to considerable increase in reactivity. No statement concerning the probability of such an accident scenario can be found.
  
  
• The reliability of high current accelerators will have to be improved (200 uncontrolled beam losses per week are present state of the art). Of course the high current (of the order of 20 mA) has still to be demonstrated (in contrast to 1.6 mA which are already reached).
  
  
• The primary coolant lead will also act as a spallation target. It will therefore be poisoned with spallation fragments. A final disposition scheme for the 10 000 t of lead is not given.
  
  
• The consequences of a break of the beam window are not yet fully analyzed. Especially the influence on reliability and availability of the overall system has to be carefully considered.
  
  
• One major aspect of the EA-concept is the use of reprocessing. This will allow in principal to achieve a reduction of long-lived radioactive waste by recycling it into new reactor fuel. The possibility to avoid the necessity of geological storage will depend strongly on the achievable separation factor (the fraction of actinides and long lived fission-products remaining in the waste stream). A comprehensive discussion of the problems related to reprocessing is beyond the scope of this paper. Of course, the possibility of reprocessing and transmutation of waste is no specific feature of the EA and has to be discussed in the broader context of nuclear energy production.
  
  
• Even if the closed fuel cycle as proposed in [Rub 97b] can be realized, a large amount of nuclear waste will have to be stored for 500 to 1000 years. It is not clear, whether near-surface storage will be sufficient for this waste problem, or if deep geologic repositories will nevertheless be necessary.

Since major promises of the Rubbia concept are higher proliferation resistance and better safety features due to the subcritical mode of operation, these aspects should be reviewed more extensively.

Military usage of Uranium-233:

A major problem concerning the civil use of nuclear energy is the production of fissile material and their possible use for nuclear weapons. This is especially true in the case of a closed fuel cycle with the appearance of separated fissile material, which is directly accessible for military use. Therefore, one has to pay special attention to the question of weapon usability of materials occurring within the process of reprocessing and fuel manufacturing.

This is also stated by the CERN-group (see point (3) of the promises). They are therefore investigating a possible diversion, stating:

"Proliferating uses of the fuel are further prevented by the fact that the fissile uranium mixture in the core is heavily contaminated by strong gamma-emitter Thallium-208 which is part of the decay chain of ²³²U and by the fact that the EA produces a negligible amount of Plutonium. As shown later on, a rudimentary bomb built starting with EA fuel, in absence of isotopic separation, will be most impractical and essentially impossible to use or hide". [Rub 97b, p. 194].

This argument is confirmed later on (p. 223 f): the required uranium for a bare critical mass (concerning the isotopic uranium composition at a discharge of 150 GWd/t) is estimated to be 28 kg. Therefore the radiation dose due to 30 kg of uranium containing 1000 ppm of ²³²U is calculated to be 36 S/hour, corresponding to 50 % lethal dose after 10 minutes exposure to the bare critical mass.¹⁰

In another paper focusing on the use of the EA as a waste burner [Rub 97a] the fuel will consist of an initial amount of plutonium from spent LWR fuel, so it will also contain a certain amount of ²³⁸U, due to the fact that during reprocessing of spent LWR fuel to extract plutonium some uranium is remaining in the plutonium stream. Therefore it is stated:

"However this uranium will mix with the ²³³U bred during the cycle, which is about 170 kg/year. As a result the produced uranium will be "denatured'' to an isotopic mixture with 63 % of fissile ²³³U, which excludes military diversions." [Rub 97a, p. 22 f]

Nonetheless these arguments will have to be reviewed carefully.

Of course, the production of plutonium in a thorium fuel is drastically reduced in comparison to ordinary uranium fuel. Therefore, the discussion has to concentrate on the produced ²³³U, which will serve the purpose of the fissile material in a thorium fuel cycle.

In table 1, numbers for critical masses of various isotopic mixtures of weapons-grade materials are given. The calculations have been performed using the Monte-Carlo Neutron transport code MCNP4A [Bri 93] for a sphere of the specified material surrounded with 10 cm of natural uranium as reflector. Even for the worst case of 40 % of ²³⁸U and 60 % of ²³³U, the critical mass is below 16 kg, therefore still comparable to highly enriched uranium.

The amount of ²³²U in the spent fuel, responsible for the formation of the gamma-emitter ²⁰⁸Tl, is strongly dependent on the final burn-up, the maximum concentration being achieved later than the equilibrium concentration of ²³³U. Due to the highly heterogeneous burn-up of the fuel, there will also be fuel elements with considerably lower concentration of ²³²U, thus reducing a potential radiation dose. In combination with a lower critical mass, this may reduce the calculated dose of 36 S/hour by an order of magnitude. Furthermore, since the uranium stemming from reprocessing initially contains no ²⁰⁸Tl, handling of the material is possible with a far lower radiation of the material during the first weeks or even months.¹¹ Up to now it seems unclear, if the finally achieved radiation barrier will be sufficient to make the assembly and use of an already built warhead impractical for technical reasons. Especially in combination with other very favorable properties of ²³³U (extremely small neutron background), a careful investigation of the overall usefulness of it as a weapons material will have to be undertaken.

 

		Reflected sphere
Material		Radius	Crit. mass	Mass with reflector
233U:	100 %	4.28 cm	6.13 kg	231.6 kg
233U:	60 %	5.81 cm	15.60 kg	315.5 kg
238U	40 %
235U:	95 %	6.25 cm	19.23 kg	341.3 kg
238U:	5 %
WPu		3.82 cm	4.64 kg	210.8 kg

Table 1: Critical masses of different nuclear weapons materials. Monte carlo calculations. WPu means weapons-grade plutonium with more than 94 % ²³⁹Pu

What level of subcriticality is sufficient/economic?

One major feature of accelerator driven reactors is the fact, that they are operated in a subcritical state. This has several consequences, which will influence the choice of the level of subcriticality r.

First of all, the subcritical mode of operation offers considerable safety margins in the case of sudden reactivity insertions into the system. As long as these will not lead to a supercritical configuration, no dramatic power excursion with severe consequences has to be faced.

Especially for the case of a high transuranic inventory, for example if the EA is operated as a waste incinerator, the margin of delayed neutrons will be relatively low, making the control of a critical reactor more difficult. In this case the additional amount of subcriticality might be regarded as a necessary improvement in reactor safety.

But apart from this beneficial effect of subcriticality, other consequences will also have to be taken into account. For a given neutron source strength S, which is directly proportional to the accelerator current, the resulting thermal power P produced in the core is given by: P = S/r

Therefore the amount of subcriticality determines the necessary accelerator strength to achieve a certain reactor power.¹²For economical reasons, the driving accelerator should be as small as possible, thereby reducing capital costs and the required amount of electrical energy. This will lead to a strong incentive to make the reactor as nearly critical as possible.

Another aspect is the influence of k_eff on the resulting neutron flux distribution, which is proportional to the local power density. While in a critical reactor the flux distribution is relatively flat, in a subcritical arrangement the flux will decrease with the radial distance from the source. For low k_eff this decrease will be exponential, therefore leading to strong power peaking in the center of the core, possibly resulting in a limitation of the total power of the system. If no fuel shuffling is taken into consideration, this effect will also lead to a very heterogeneous burn-up of the fuel, which may pose problems to fuel integrity and is undesirable for economical reasons.

To highlight the fact, that the resulting dilemma is also present in the papers concerning the EA, some quotation shall be given:

"Our proposed procedure is definitely attractive for many reasons, amongst which: ... (3) from the safety point of view, since the device is not "critical'', namely it keeps the waste at all times in conditions k £ 0.95 which are essentially the same as the ones in the projected geologic repository." [Rub 97a, p. 4].

"As we shall see the same condition k £ 0.95 can be preserved at all times during the incineration process in the EA."[Rub 97a, p. 9].

"The actual choice of the operating k for the EA depends on the fraction of energy recirculated through the Accelerator. ... In the paper we use k = 0.97, that we believe safe enough". [Rub 97a, p. 9], footnote to last quotation.

While a value of k £ 0.95 is stated to be attractive from the safety point of view, a higher value is chosen for economical reasons. If the given reference value of k = 0.97 is compared with the actual values of k during burn-up as given in [Rub 97], it is evident that during at least one half of the whole life cycle of the fuel the value of k is still higher than 0.97.

Since no strong criteria for the determination of an acceptable level of subcriticality seem to exist, this point will therefore have to be thoroughly clarified.

Conclusions

The main feature specific to the EA and any accelerator driven reactor is the subcritical mode of operation. Although this may result in certain safety advantages, other accidental scenarios related to a sudden increase in beam power will have to be carefully investigated before the overall effect of subcriticality can be judged. Since the level of subcriticality strongly influences the overall economy of the system, a thorough determination of the necessary level of subcriticality has to be undertaken.

Other safety arguments like the removal of heat by natural convection or the use of lead instead of sodium as a primary coolant are not specific to the EA and resulting problems are not yet fully specified or solved.

The assertion of higher proliferation resistance is mainly based on the fact, that ²³³U is bred as fissile material instead of plutonium in the case of an uranium fuel cycle. Nonetheless the weapon-usability of ²³³U is at least not to be excluded. The emergence of separated weapons-grade uranium due to the necessity of reprocessing therefore poses comparable proliferation problems as those related to other separated weapons-grade materials in the civilian fuel cycle.

Notes

1. This is a slightly modified version of a paper originally presented at the Tenth International Summer Symposium on Science and World Affairs, MIT, Boston, July 14—21, 1998.

2. It is understood that the discussion about the role and consequences of nuclear energy production is by far beyond the scope of this paper.

3. Of course, several specific design parameters may change (and have already), therefore some of the specific arguments given below may have to be adjusted according to changes in design.

4. Due to changes in the level of subcriticality of the device, the accelerator must be able to produce up to 20 mA*GeV.

5. The multiplication factor k_eff describes the development of the systems neutron population. It is defined as the ratio of two successive neutron generations. Therefore you have a critical reactor, if the value of k_eff is exactly equal one. Below this value, a system is subcritical, the number of neutrons in each neutron generation is smaller than in the preceding one by a factor of k_eff.

6. This means that approximately 5 % of the produced electric energy will have to be fed back to the accelerator, assuming a 45% thermodynamical efficiency due to the high operating temperature of the EA.

7. The initial design value for the height of the vessel was 30 meters, later reduced to 15 meters. The diameter of the vessel will be 6 meters.

8. This is stated to be a step to increase the overall proliferation resistance of the EA, since no handling of the fuel is necessary during normal reactor operation, thereby simplifying safeguards.

9. It has to be mentioned, that different concepts for the fuel-cycle of the EA are discussed, some of them also mixing the fuel-cycle of the EA with that of ordinary light water reactors. Insofar, no general discussion of the fuel-cycle of the EA is possible.

10. The ²⁰⁸Tl is built up by the decay of ²³²U and reaches the asymptotic concentration after approximately 1000 days.

11. After 200 days, the ²⁰⁸Tl-concentration is still less than 20% of its maximum value.

12. Since the subcriticality r is related to the multiplication coefficient k by r = (k-1)/k, already small changes in k will lead to considerable changes in the produced power of the system. Decreasing k from 0.97 to 0.95 will reduce the power to 59 % of its initial value.

References

[Art 95]: Arthur, E.; Rodriguez, A.; Schriber, S. O. (eds.): Proceedings of the First International Conference on Accelerator-Driven Transmutation Technologies and Applications, Las Vegas, 25—29 July 1994, AIP Press, Woodbury, NY, 1995.

[Bri 93]: Briesmeister, J. F. (ed.): MCNP — A General Monte Carlo N—Particle Transport Code. Version 4A, LA-12625-M, Los Alamos National Laboratory, November 1993.

[Con 97]: Condé, H. (ed.): Proceedings of the Second International Conference on Accelerator-Driven Transmutation Technologies and Applications, Kalmar, Sweden, 3—7 June 1996, Uppsala University, 1997.

[HSK 97]: On the Feasibility and Safety of the Rubbia Energy Amplifier (EA) with exclusion of non-proliferation aspects and use of military plutonium, Paper by a Group of Experts from the Swiss Nuclear Safety Inspectorate (HSK) assisted by Paul Scherrer Institute (PSI), 30 June 1997.

[Rub 94]: Rubbia, C.: A High Gain Energy Amplifier Operated with Fast Neutrons, in [Art 95].

[Rub 97a]: Rubbia, C. et al.: Fast Neutron Incineration in the Energy Amplifier as Alternative to Geologic Storage: the Case of Spain, CERN/LHC/97-01, 1997.

[Rub 97b]: Rubbia, C. et al.: Cern-Group Conceptual Design of a Fast Neutron Operated High Power Energy Amplifier, in: Accelerator driven systems: Energy generation and transmutation of nuclear waste. Status report, IAEA-TECDOC-985, November 1997.

Christoph Pistner is researcher at IANUS, Technical University Darmstadt, Hochschulstraße 10, 64289 Darmstadt, Germany, tel +49 -6151 -163016, fax 166039, email:christoph.pistner@physik.tu-darmstadt.de