A new non-proliferation assessment tool

Proliferation resistance is defined by IAEA as ‘that characteristic of a nuclear energy system that impedes the diversion or undeclared production of nuclear material or misuse of technology by states in order to acquire nuclear weapons or other nuclear explosive devices’ [1]. In the context of this report, physical protection refers to those features of the nuclear system that provide intrinsic protective barriers that help prevent nuclear materials being accessed by a terrorist group. This is the same as used by Generation IV International Forum (GIF) [2], which define the threats under physical protection as originating from ‘a sub-national group or other non-host state adversary’ and involving ‘either theft or sabotage’.

International efforts to stimulate the development of advanced nuclear systems, such as the GIF and the IAEA International Project on Innovative Nuclear Reactors and Fuel Cycles (INPRO) project, have identified the need to strengthen features that provide inherent proliferation resistance and physical protection (PRPP) as one of their main goals. Current reactor systems and their associated fuel cycles were not primarily designed for inherent PRPP. In some instances they rely on institutional measures to control access to nuclear materials to prevent diversion or theft. It is not the intention that inherent physical characteristics would at any time lead to nuclear safeguards and security being diminished in any way. Instead, the aim is to be able to demonstrate inherent PRPP by design (such as avoiding the production or separation of high-quality fissile material) and thus to help increase transparency and consistency in decisions on system development and ultimate selection. For the purpose of evaluating options against PRPP, no credit is assumed for safeguards and security measures, even though these will always be applied in full in practice.

The UK National Nuclear Laboratory (NNL) has been pro-active in developing a PRPP assessment methodology in support of UK requirements in this area. Although PRPP assessment methods were developed primarily to guide decision-making on advanced reactor and fuel cycle R&D, they can also be applied in other contexts, such as the management of historic UK liabilities.

Nuclear reactors and nuclear weapons both rely on fissile materials such as U-235, Pu-239 or U-233. These are nuclides with a strong propensity to undergo nuclear fission on interacting with fast or thermal neutrons. All reactors need to be supplied with fissile material (in fresh fuel) and there is always some residual fissile material (in the spent fuel). In principle, fissile materials could be diverted from fresh fuel or spent fuel and misused for making nuclear weapons. Therein lies the risk of nuclear weapons proliferation, which an ideal reactor system and its fuel cycle would be designed to minimise. This could be achieved by avoiding the use of high-quality fissile materials, avoiding the separation of pure fissile materials in the fuel cycle, or by building into the design of a reactor system inherent features which would make it difficult to isolate high-quality fissile materials, or which would help ensure the early detection of any attempts to divert such materials.

Given that inherent PRPP is one of the major goals of GIF and INPRO, it is important to have some objective means of assessing the relative strengths and weaknesses of different systems and their associated fuel cycles. Many different methods have been developed by the international community for this purpose. None can claim to be perfect; each method has strengths and weaknesses (see below for discussion). But since all of them involve a greater or lesser degree of subjectivity, none can be regarded as completely objective. Nevertheless, despite their limitations, PRPP assessment methods do have a useful role to play in advanced reactor and fuel cycle development. The benefit of such analysis often comes not just in the final results of the assessment, but also out of the discussions and debate that take place in reaching the consensus scoring.

Proliferation risks

The scope of the threats considered in this paper is restricted to the diversion or theft of fissile materials. In the case of diversion, the threat is postulated to come from state-sponsored diversion of fissile material from safeguarded civil nuclear facilities. The eventual aim would be to divert sufficient fissile material to assemble into one or more viable fission-based explosive devices. The intended purpose of such devices might range from a reliable high-yield strategic or tactical weapon to a low-yield device. For a high-yield device, high quality fissile material would be considered preferable, but it is widely acknowledged that poor-quality fissile material is potentially usable, albeit with additional technical difficulties. For a low-yield device, fissile quality is not considered a limitation and any fissile material would be considered to pose a threat. At the limit, mere possession of diverted fissile material without any attempt to assemble a fissile explosion device might suffice to pose a serious threat.

In the case of theft, the threat is postulated to come from the theft of fissile materials from safeguarded civil nuclear facilities by a sub-national organisation. The scenario assumes that nuclear security measures designed to protect against theft have been circumvented. A sub-national organisation might then intend the nuclear material to be used for the assembly of a viable fissile explosive device. A high-yield device is likely to be beyond the capabilities of a sub-national group, but cannot be ruled out. A low-yield device is more likely. Again, mere possession of fissile material would in itself constitute a serious threat.

All fission reactors and their associated fuel cycle pose some degree of proliferation risk. The front-end and back-end of the fuel cycle feature the greatest risks. Freshly-manufactured natural uranium and low enriched uranium fuels used in current commercial reactors are considered to pose minimal proliferation risk because the fissile material is not in weapons-useable form and would need to be enriched to much higher levels to pose a viable threat. Historically, many research reactor fuels used high enriched uranium (HEU), with enrichments over 20 weight percent U-235, which is weapons-useable. In response to international pressure, most research reactors have now been converted to use lower initial enrichments, which lessens the extent to which institutional controls are necessary to protect the fuel. It is worth noting that uranium enrichment technology is widely regarded as posing one of the major proliferation threats.

At the back-end of the fuel cycle, spent nuclear fuel contains residual quantities of fissile materials. Spent uranium fuel from commercial reactors typically contains up to 1.0 weight percent of residual U-235 and a similar concentration of plutonium generated by neutron captures in U-238 (this is the fertile conversion mechanism, whereby U-238 is converted to fissionable Pu-239 by neutron captures). The plutonium in spent nuclear fuel can be separated by reprocessing and thereby generates a source of potentially weapons-useable fissile material. The physical bulk of spent nuclear fuel, combined with the need to chemically separate the plutonium and the presence of a strong irradiation field, provides a certain degree of self-protection. Separating plutonium in reprocessing makes the fissile material more accessible, which implies that more reliance must be placed on institutional control measures (nuclear accountancy, safeguards and security) to guard against diversion or theft. Recycling the plutonium as mixed oxide (MOX) fuel is beneficial because, not only is a proportion of the plutonium actually destroyed (approximately 25%), but in addition, the plutonium becomes difficult to access once the MOX fuel has been loaded and irradiated in the reactor.

Similar questions as to the balance between inherent protection and institutional control occur throughout the nuclear fuel cycle. Both GIF and INPRO have expressed a preference that future nuclear systems should lean more towards the former. This preference brings with it a clear need to be able to demonstrate that a system has a high degree of inherent PRPP. This is the reason why there has been increasing interest in PRPP assessment methods. The aim is to be able to compare the intrinsic PRPP characteristics of different reactor and fuel cycle systems in a transparent and consistent way, in order to be able to demonstrate that PRPP aspects have been addressed in the selection process and identify those areas of the fuel cycle where there is potential scope for improvement.

PRPP assessment methods have a broader role to play in the justification process for decisions involving the UK nuclear fuel cycle. It is a formal European Union requirement that any practice involving radioactive materials needs to undergo a justification process that demonstrates clearly that the benefits outweigh the detriments. This is a broad-ranging process that addresses relevant aspects including safety, environmental impact and radiological impact. PRPP is regarded as one of the aspects that needs to be included.

Technical Issues

Quality

Assessing the inherent PRPP characteristics of a reactor system and its associated fuel cycle is a complex inter-disciplinary task, but it can be broken down into two major technical areas, fissile material quality and fissile material accessibility.

The physical properties of fissile material determine whether it is weapons-useable. There are five main factors.

1. Neutron multiplication factor. To be weapons-useable, fissile material should have a neutron multiplication factor (the ratio of the number of neutrons generated to the number consumed in atomic collisions) well in excess of 1.0 when assembled in a critical geometry (that is, a neutron collision would generate in excess of one new neutron for every one that started the collision). A high multiplication factor reduces the number of neutron generations needed to achieve a full yield and thereby strongly affects the viability of a fission device. High enriched uranium (HEU, normally defined as uranium enriched in U-235 to >20 weight percent), plutonium with a high content of Pu-239, and U-233 all satisfy this requirement.

2. Fissile mass. The fissile mass determines how compact a fission device needs to be and is also a major factor in determining its viability. HEU, plutonium with a high content of Pu-239, or U-233 all have low critical masses.

3. Spontaneous neutron source. Emissions of spontaneous neutrons can cause the premature initiation of neutron multiplication. HEU, Pu-239 and U-233 all have very low spontaneous neutron emissions. The unavoidable presence of Pu-240 at varying concentrations in plutonium introduces a large spontaneous neutron source that may affect the reliability of achieving high yields.

4. Heat output. The presence of heat-producing nuclides such as Pu-238 will cause the temperature of the components of a fission device to increase, possibly beyond workable limits. Heat output is a major complicating factor that may in some circumstances demand that assembly of a fission device be carried out very shortly before use. HEU and U-233 have low heat outputs, while reactor-grade plutonium has a significant heat output because of the presence of Pu-238.

5. Dose rate. The radiological dose rate of the nuclear material is an important consideration in fabricating a fissile device. The presence of gamma emitters in some fissile materials may require radiological protection measures. HEU has a low radiological dose, while plutonium and uranium-233 can both have significant radiological emissions, depending on their isotopic makeup.

To protect against diversion, it is preferable to have poor-quality fissile materials in the civil nuclear fuel cycle. Material with a high critical mass, high spontaneous neutron source and high heat output would complicate weaponization and lower the probability of a ‘successful’ outcome. However, it is widely agreed that, within reason, a state with access to sophisticated weapons expertise could overcome any such difficulties and all plutonium is therefore considered to pose some degree of threat. Because of the obvious sensitivities, the detailed assessments are not available for open discussion.

For the theft scenario, the physical properties of the fissile material are largely irrelevant, because the mere possession of fissile material or the assembly of a low-yield device already poses an unacceptable threat.

Accessibility

Spent nuclear fuel typically contains just 1 or 2 weight percent of fissile material (uranium plus plutonium primarily) dispersed in the fuel matrix. Separating the fissile material in reprocessing requires shielded handling of the spent fuel assemblies, a task normally carried out in a pond with several metres of water covering the top of the assembly. The fuel assembly then needs to be sheared mechanically and then dissolved in acid. The fissile materials (uranium and plutonium) are separated from the fission products, other heavy elements, and from each other and then purified to remove all residual radioactive contaminants.

All these steps represent physical barriers to accessibility. A would-be clandestine proliferator needs to construct and operate complicated large scale industrial facilities in order to be able to accomplish them. There are major technical and cost barriers to building such a facility, and they would also require time to overcome. The existence of such facilities, and the various physical, chemical and radiological signatures associated with operating them, put the entire process at risk of detection.

Spent fuel can therefore be regarded as having a high degree of inherent PRPP. GIF and INPRO both recognise the need to increase the inherent PRPP of reprocessing by ensuring that, where possible, the separation of pure fissile material for recycle is avoided. A possibility being considered includes the production of a mix of uranium and plutonium rather than pure plutonium.

In NNL’s experience, fissile material accessibility has been the main determinant of PRPP. In future, there needs to be increased attention on material accessibility in designing reactors and fuel cycles to maximise inherent PRPP.

International PRPP methodologies

In the past 10 years four major international research projects have addressed PRPP either as their principal focus or as part of the broader requirements of the development of future nuclear systems. All four of these have developed what are loosely referred to as PRPP methodologies. However, these so-called methodologies could perhaps be more accurately regarded as providing general guidelines or frameworks for PRPP assessment, rather than prescriptive methodologies.

The U.S. Department of Energy (DOE) Office of Nuclear Energy, Science, and Technology and Nuclear Energy Research Advisory Committee (NERAC) established a special Task Force in 1999 to review the technological opportunities to improve PRPP for global civilian nuclear power systems (TOPS) [3]. The TOPS report was one of the first international studies to consider how civil nuclear systems design might be directed towards enhanced PRPP. The PRPP approach adopted was entirely qualitative, based on breaking down the fuel cycle into stages and using a tabular method whereby the barriers applicable at each stage (both institutional barriers and inherent technical barriers) are listed and ranked according to a five-point scale as follows: ineffective, low, medium, high or very high. The tables so constructed can then be used to identify priority areas for maximising overall effectiveness. Its purpose was not so much to arrive at a definitive assessment, which was considered unrealistic at that time, but to use the comparative analysis to highlight the key technical questions that a fully developed methodology would need to address.

Japan Atomic Energy Agency (JAEA) developed the TOPS approach further [4, 5] to introduce an element of consensus. It uses the TOPS tabular approach, but the barrier scoring is carried out by a panel of experts whose scores are then averaged to arrive at a numerical ranking. This is an attempt to make the process less affected by individual subjectivity. Although the outcome is a numerical score, the method is still largely qualitative and therefore unable to provide reliable relative rankings between options or sensitivities. It is, however, generally applicable to any stage of the fuel cycle, and like TOPS can highlight where there may be areas for potential improvement.

The IAEA’s INPRO has developed a set of guidelines to assist developers of new nuclear systems with all aspects of system design, including PRPP [6]. The aim of INPRO is essentially to establish best practice for nuclear system designers to follow when developing new reactors and their associated fuel cycles. For all of the technical areas covered by INPRO, a set of basic principles and user requirements are defined [6]. For PRPP the basic principle is:

“Proliferation resistance intrinsic features and extrinsic measures shall be implemented throughout the full lifecycle for innovative nuclear energy systems to help ensure that innovative nuclear systems will continue to be an unattractive means to acquire fissile material for a nuclear weapons program. Both intrinsic features and extrinsic measures are essential, and neither shall be considered sufficient by itself.”

The Generation IV Forum (GIF) was set up to encourage the development of new reactor systems and, in a similar way to INPRO, has developed best practices for designers to follow. One of the areas covered is PRPP, where an assessment methodology has been developed [7]. This has some similarities to the INPRO methodology, being largely qualitative and based on prescriptive categorisations, but is currently more developed, providing a semi-quantitative outcome and is perhaps somewhat less subjective. It can also be used to analyse different threat scenarios.

The Gen IV PRPP methodology is a rigorous process that is intended to be carried out by a large team of people, possibly from different organisations. Once the threat pathway has been defined, the Gen IV PRPP methodology involves analysing the proliferation risk based on a set of six metrics, each of which assigns a nuclear system into one of five categories (very low; low, medium, high or very high). The six metrics are technical difficulty, proliferation cost, proliferation time, fissile material type, detection probability and detection resource efficiency, all of which are assigned into one of five pre-defined categories, which simplifies the assessment process.

Assigning scores against each of the six metrics is necessarily a subjective activity, though the degree of subjectivity is minimised by providing clear guidelines as to which category is applicable given the materials in the fuel cycle. For example, the cost of implementing a proliferation pathway is arbitrarily set to a fraction of a state’s military resource budget. The five categories equate to 0-5%, 5-25%, 25-75%, 75-100% and >100% of the country’s annual defence budget, respectively.

The results of the Gen IV PRPP methodology can be presented in tabular form and/or graphically, with the metrics in rows and scores in columns. The table and graph provide a qualitative guide to comparing different proliferation pathways for a given reactor/fuel cycle. The method does not use weighting functions to aggregate the different metrics; no single overall numerical score is provided.

A large number of PRPP methodologies have been developed that have been influenced by and designed to be consistent with one or more of the frameworks above. Their common distinguishing feature is that they aim to combine all the different metrics to provide a single quantitative figure of merit with which to compare different systems. In principle, this also allows sensitivity and uncertainty analysis to be performed, which is difficult to do in any meaningful way with qualitative approaches.

Care is required, however because although these quantitative methods may give the appearance of being rigorously objective, in many cases the process still involves a degree of subjectivity. Three examples are summarised below.

(a) Multi-Attribute Utility Analysis (MAUA) is a long standing method that has been used to aggregate the assessment of multiple attributes. For PRPP analysis, MAUA has been developed most fully by Texas A&M University (TAMU) [8,9]. MAUA is useful because it generates a single figure of merit for the entire system, but is limited because the weights and utility functions used in the method are subjective, and it is difficult to know how much the overall result is affected by this subjectivity. MAUA can be used to determine the time dependence of proliferation risk; any time dependence of an individual attribute (for example, due to radioactive decay) affects the individual utility functions and propagates to the overall utility function.

(b) Markovian methods can be used for the analysis of a specific system against a specific threat scenario. It relies on constructing a Markov chain to represent the pathways of normal flow of fissile materials through the fuel cycle, and postulated diversion pathways by which a proliferator attempts to acquire weapons usable fissile material. The method has been developed by Brookhaven National Laboratory [10,11] and is designed to calculate the probability of detection of covert diversion activities or, conversely, the probability of successful diversion. Provided that the detection probabilities are evaluated rigorously, it is potentially very useful for assessing the effectiveness of safeguards measures, and in particular of highlighting where potential vulnerabilities may lie.

(c) Risk-Informed Probabilistic Analysis (RIPA) is a method based on probabilistic risk assessment (PRA). RIPA is designed to determine the most probable path towards acquisition of a nuclear weapon and thus to show how the fuel cycle can best be modified to minimise the probability of proliferation success [12]. RIPA accounts for the time and cost of each stage in the proliferation chain (evaluated taking account of the proliferator’s available resources), the probability of detection, the probability of technical success and the materials throughput. RIPA gives an objective, quantitative assessment of the proliferation risk and in principle can be used to carry out sensitivity analyses. It does not require the use of weighting factors as in MAUA, eliminating this source of subjectivity. However, there are still subjective elements, because not all the elements of the analysis may necessarily be quantifiable. For example, the probability of detection of different proliferation options might need to be estimated by an expert panel, which would introduce a degree of subjectivity.

NNL has reviewed the potential suitability of the various approaches described above for application in the UK and carried out a comparative assessment. The conclusion of the assessment is that none of these approaches were ideally suited for the applications envisaged in the UK, although there were elements in the INPRO and GIF methodologies which were potentially useful. The assessment established the limitations of these approaches.

1. Many of the approaches (TOPS, JAEA and INPRO) are qualitative in nature and therefore not able to provide sensitivity analysis or uncertainty analysis.

2. Only some of the approaches (INPRO, MAUA and the Markovian method) incorporate time dependence.

3. None of the methods is entirely free of subjectivity. However, some of the methods involve a higher degree of subjectivity than others.

4. Some of the methods are better suited to assessing a specific proliferation scenario and are less well-suited for assessing the generic risk posed by a particular system.

5. The MAUA, Markovian and RIPA methods are not straightforward to implement; they require specialist software and sometimes require very time-consuming and demanding inputs from users.

The survey reveals that the TOPS, JAEA and INPRO methodologies all have limitations, especially a failure to provide quantitative results. This is partly because they are really high-level frameworks that are intended to direct more detailed work. Also, the current version of the INPRO methodology is not complete, although the finished version may address some of its limitations.

NNL therefore decided to develop its own approach, incorporating some of the best features of the available methodologies and attempting to avoid their limitations as far as possible. NNL’s approach is designed to be flexible enough to be applicable to the various different UK applications and easy enough to allow relatively fast turnaround on application. More specifically, seven user requirements have driven the development of NNL’s PRPP assessment methods:

1. To be as objective as possible

2. To provide a numerical relative ranking of different reactor and fuels cycle systems (though there is no requirement for the numerical ranking to have any absolute quantitative meaning)

3. To be applicable to any reactor and fuel cycle system and any proliferation pathway

4. To be sufficiently general to be able to incorporate different approaches to treating the various metrics adopted and to allow easy inter-comparison

5. To be simple to implement and clear to display graphically

6. To use input parameters preferably that are easily obtainable without requiring complex shielding or critical-mass calculations

7. To be usable without reliance on knowledge of sensitive site-specific information, such as details of the security measures in place at a nuclear plant. Such information is understandably kept confidential and its inclusion would not allow the methodology to be used in open publications.

NNL’s methodology is based on calculating a ‘utility function’ that represents how useful a nuclear fuel cycle might be to a potential proliferator and in which the multiple parameters that characterise the fuel cycle are captured. The utility function is the product of a ‘value function’ that measures the mass of fissile material in the fuel cycle and its fissile quality, and an ‘access function’ that measures how accessible the fissile material is. These functions are generally applicable to both the front-end and the back-end of the fuel cycle. Since it is the access function that is the main controlling parameter, it is informative to plot the utility function as a function of the access function, as this visualises where the inherent PRPP attributes of a system come from. The U-A methodology draws its name from this practice. The U-A methodology incorporates aspects of MAUA, but in a manner which is much simpler and easier to implement. The U-A methodology also uses the GIF metrics to determine the value and access functions, so that it is automatically links in with the GIF framework. Although NNL’s methodology addresses all of these user requirements, it does not completely avoid subjectivity, as the assessments are completed by expert individuals, or groups of individuals.

To provide confidence, NNL has benchmarked its U-A methodology against the various international approaches discussed earlier [13]. Applied to analysing different fuel cycles, the U-A methodology gives relative rankings that agree very well with other approaches. One of the main conclusions of the benchmarking was that there was a high degree of correspondence between the various different methods, which adds confidence that the results from the different approaches are meaningful even though none of the methods can give an absolute measure of PRPP.

NNL has applied the U-A methodology to a number of different applications in the UK and there are plans to extend its application further.

Conclusions

Having reviewed international PRPP assessment methods and having developed and applied its own U-A method to UK fuel cycle options, NNL has drawn eight conclusions.

• Underlying the entire approach to PRPP assessment is an implicit acceptance that no option is completely free of proliferation risk. The method has no formal role in security assessment and in determining the measures that are required to safeguard nuclear material. However, the method can identify options which might be preferred because, for example, a favoured option might be more reliant on intrinsic rather than extrinsic measures.
• PRPP assessment methods are only able to indicate relative risks and it is not feasible to quantify the absolute risk of proliferation, nor is it sensible to attempt to define any kind of go/no-go criterion.
• Accepting that there is no perfect method for comparing the proliferation risk posed by different options, there is still merit for the UK in applying the methodology and presenting and recording the results as part of due process in a consultation system that is intended to be consistent, open and transparent.
• Given the shortcomings in the international methods developed to date, NNL has learnt from international best practice and developed its own approach, which is designed to be flexible enough to be applicable to the various different UK and international applications and easy enough to allow relatively fast turnaround on application.
• NNL’s methodology demonstrates a simple and practical means to combine PRPP metrics, such as those defined by GIF, into a single figure of merit that can be used to rank different fuel cycle options. Using the GIF PRPP metrics as the starting point automatically ties the NNL methodology to international practice. NNL’s methodology has been successfully benchmarked against international best practice.
• The need to assess options against the GIF PRPP metrics has, of itself, proved very valuable, automatically focusing the discussion to ask the right questions and facilitating the agreement of a consensus view when completing the analyses. It is during the discussion and debate within a group of experts where the real value is added and the methodology facilitates that process. This approach is actively encouraged.
• Successful applications of the NNL methodology to date include an assessment of the inherent PRPP of options to manage the UK’s stocks of separated plutonium, immobilised waste forms options, spent fuel management and reprocessing options. These applications of the NNL methodology are sufficient to demonstrate that it is possible to rank different fuel cycle options relative to each other in a manner which is transparent, and to a large extent objective and reproducible, to assist policy and decision- makers from a PRPP perspective.
• NNL’s experience is that fissile material accessibility is the key determinant of inherent proliferation resistance. Although fissile material quality is important, it plays a secondary role.

Further, more detailed, information regarding NNL’s PRPP methodology, its comparison with international alternatives and its application to the UK’s plutonium management options is available in published literature [13,14,15], as is further background information regarding how NNL supported the recent Royal Society project on nuclear non-proliferation [16].

Kevin Hesketh, Senior Fellow, National Nuclear Laboratory, 5th Floor, Chadwick House, Warrington Road, Birchwood Park, Warrington, Cheshire WA3 6AE UK

This article was originally published in the August 2012 issue of Nuclear Engineering International