FHR, HTGR, and MSR Pebble-Bed Reactors with Multiple Pebble Sizes for Fuel Management and Coolant Cleanup

Abstract

Three reactor types can be designed with pebbles (carbon spheres) as the reactor core: the pebble-bed high-temperature gas-cooled reactor (PB-HTGR), the pebble-bed fluoride-salt-cooled high-temperature reactor (PB-FHR), and the thermal-spectrum molten salt reactor (MSR) with fuel dissolved in coolant. In the HTGR and FHR, the pebbles are fuel (coated-particle fuel) and moderator (graphite). In a MSR the pebbles would be the moderator (no fuel). Recent advances enable prediction and modeling of pebble beds with two or more sizes of pebbles.

This may enable the use of pebble beds with multiple size pebbles that create new options. A second smaller size of HTGR/FHR fuel pebble that fills some of the space between the regular pebbles can increase the power output for the same size reactor. For the FHR the second pebble size would reduce inventory of expensive coolant and may widen choices of salt coolants. In an HTGR or FHR, smaller pebbles with high actinide loadings and high heat transfer rates could be used to burn actinides while the larger pebbles are the driver fuel. Multiple pebble sizes in MSRs may enable varying the carbon-to-fuel ratio to optimize the neutron spectrum over time to more efficiently utilize the fuel and allow easy replacement of moderator. The smaller pebbles with no fuel and a high surface-to-volume ratio could be designed to remove (1) HTGR/FHR/MSR tritium from the coolant and (2) noble metal fission products and potentially other impurities in MSRs. We examine the potential incentives for pebble beds with multiple size pebbles. With the tools now available to quantify pebble-bed behavior with multiple size pebbles, the next step is to begin to quantify benefits and limitations for different applications of pebble-bed reactors with multiple sizes of pebbles.

Keywords:

• Fluoride-salt-cooled high-temperature reactor
• molten salt reactor
• high-temperature gas-cooled reactor
• pebble-bed reactor

I. INTRODUCTION

Three reactor types can be designed with carbon spheres (pebbles) as part of the reactor core. The carbon acts as a neutron moderator in all cases. In the pebble-bed high-temperature gas-cooled reactor (PB-HTGR) and the pebble-bed fluoride-salt-cooled high-temperature reactor (PB-FHR) the pebbles contain fuel. The pebbles flow through the reactor core over time with partly burnt pebbles recycled again through the reactor core and fully burnt pebbles removed as spent nuclear fuel (SNF) and replaced with fresh fuel pebbles. Pebble size is uniform to enable prediction of behavior during reactor operations and to assure safety. One could use two or three sizes of pebbles where the smaller pebbles fill the void spaces between the larger pebbles, but this has not been done because of the inability to predict pebble flow, coolant pressure drop, and other characteristics. Thermal-spectrum molten salt reactors (MSRs) with pebble-bed moderators have not been considered because the fuel salt–to-moderator ratio is fixed in a pebble bed with a single size of pebble. About 63% of the core would be pebbles. The volume fraction of pebbles is less than the ideal packing fraction for spheres and is dependent upon multiple factors such as the pebble-to–bed diameter ratio.

Technology advances may now allow us to predict the behavior of different sizes of pebbles flowing through the reactor core. This raises the questions: Are there benefits to multiple pebble sizes? How large are the benefits? What are the disadvantages? Multisize pebbles (1) enable a larger volume fraction of the reactor core to be filled with solid pebbles and (2) result in much larger mass and heat transfer associated with the smaller pebbles than with the larger pebbles due to the higher surface-to-volume ratio. We examine the new ability to predict the behavior of pebble beds with multiple size pebbles and the potential implications of this technology. We do not undertake detailed analysis of these options but rather lay out a roadmap of options and questions.

II. PREDICTING PEBBLE-BED BEHAVIOR

Multiple single-size pebble-bed helium-cooled reactors have been built. One small test reactor is operating in China and two modular pebble-bed reactors will shortly start up in China. Pebble-bed FHRs are proposed.^1,2 Outside the nuclear industry there are pebble-bed recuperators and pebble-bed chemical reactors where the pebbles are coated with catalysts. This includes pebble beds with single and multiple size pebbles. There is massive literature^3,4 on the behavior of pebble beds. The use of pebble beds with multiple size pebbles has not been investigated in the nuclear industry because of the need to fully understand system behavior to assure safety.

The potential advantages of multisize pebbles allow a larger fuel, moderator, or other solid-to-liquid or solid-to-gas ratio in an environment with very good heat and mass transfer. The use of multiple pebble sizes depends upon the stability of multisize pebbles flowing through the reactor and our understanding of that flow for a random-packed pebble bed. This includes heat transfer for the different size pebbles and pressure drop across the core. For many of the design options, multiple size pebbles would increase pressure drop per meter of bed height but reduce the core height.

Changes in the last several years have made it possible to consider multiple sizes of pebbles for nuclear applications. First, random pebble-bed flow can now be experimentally measured. There have been large-scale pebble-bed flow experiments that tracked pebble-bed flow.⁵ More recently flow of pebbles has been measured using X-ray tomography where each pebble has a metallic pin. This allows for simultaneous following in three dimensions of the movement of each pebble, including rotation. These experimental techniques provide a method to validate advanced modeling methods that allow prediction of the behavior of flowing pebble beds with randomly packed pebbles of (1) different sizes and (2) different ratios of large to small pebbles.

Second, there are now tools to model pebble-bed flow^6–8 to answer multiple questions. This includes discrete-element simulations in realistic geometries with up to 440 000 pebbles. What is the optimum size of the smaller pebble? Will bed behavior change with bunching of small pebbles or bridging caused by the interactions of the two pebble sizes? What is the coolant fraction for different ratios of pebble sizes and numbers of each size pebble? What is the impact on heat transfer? Will the pressure drop across the column be acceptable? The suite of tools to enable realistic modeling did not exist a decade ago.

An example of a multisize pebble simulation is shown in Fig. 1 with pebbles in a cylinder (height: 78 cm, diameter: 72 cm, and volume: 0.3175 m³), small pebbles (diameter: 1 cm, volume of pebble: 5.235E–07 m³, number of small pebbles: 111 904, and total volume of small pebbles: 0.0585 m³), and large pebbles (diameter: 6 cm, volume of pebble: 1.13E–04 m³, number of large pebbles: 1676, and total volume large pebbles: 0.1895 m³). For this case the total volume of small and large pebbles is 0.2481 m³ with a packing fraction (total volume of pebbles/cylinder volume) of 0.7813. By modeling different relative sizes (two or three sizes) and numbers, the option space for random pebble beds can be mapped.

Fig. 1. Mixture of pebble sizes for described case.

One major consideration is pebble density. If the pebble densities of large and small pebbles are different, there will be separation by density, something usually to be avoided. For nuclear applications, one can design pebbles of different sizes to have the same densities to avoid separations by density. The density of graphite can be modified. If there is a mixture of fuel and nonfuel pebbles, coated particles with zirconium oxide (ZrO₂) can be added to nonfuel pebbles to match the densities of pebbles with fuel. If one wants multiple size pebbles where all the pebbles contain fuel, one can vary enrichment of different size pebbles to vary power density to fully utilize the greater surface-to-volume ratio of smaller pebbles for heat transfer.

There are other mechanisms⁹ that can result in segregation. It is the recent ability to predict behavior under different conditions that is the potentially enabling technology for using multiple pebble sizes. In parallel, other advances¹⁰ enable prediction of heat transfer and neutronics for such systems.

III. IMPLICATIONS OF MULTIPLE PEBBLE SIZES

If multiple pebble sizes can be used in HTGRs, FHRs, and MSRs, there is the potential for major economic improvements and new capabilities that these reactors do not currently have. We describe the potential applications. There has been little work to quantify either the benefits or liabilities of these options.

III.A. Higher Fuel Loadings and Smaller SNF Volumes in HTGRs and FHRs

In typical HTGRs and FHRs, all the pebbles are fuel pebbles. If one adds a second size of smaller fuel pebbles that replaces some of the coolant fraction, the power output for the same reactor core volume increases—more fuel in the same space. The smaller pebbles with higher surface-to-volume ratios and better heat transfer allow for higher power densities in the smaller pebbles. This can be achieved by higher fuel loadings or enrichments. Initial power densities in the smaller spheres will be much higher but will decrease with time, thus the benefits are not linear with enrichment. The power uprate may be greater than just the benefits from a larger volume of fuel in the reactor core. The penalty during reactor operation will be a greater coolant pressure drop across the reactor core. Such a design change involves multiple trade-offs between average core power density, pumping power, and core height.

The increased fuel packing has a significant fuel cycle benefit. The once-through fuel cycle impact is that the same waste container volume will contain more SNF with a mixture of large and small pebbles and greater total megawatt days of burnup. Historically these carbon-based fuels have had large SNF volumes relative to other reactors implying higher costs for all back-end fuel cycle operations from storage to repository disposal. Can one reduce the volume by 30% per megawatt day? In this context, these carbon-based fuels have much higher temperature limits in storage than light water reactor (LWR) SNF; thus, they do not have the same thermal constraints in storage or disposal.

III.B. Actinide Burning in HTGRs and FHRs

Various schemes have been proposed to burn actinides using thermal-spectrum reactors. Pebble-bed reactors with two pebble sizes may provide a practical method for a once-through burnup of actinides before disposal. Coated-particle fuel is capable of burnups in excess of 800 000 MWd per ton—far higher than any other fuel type. There were early programs for plutonium irradiation of coated-particle fuel¹¹ and later irradiations of high-enriched fuel that went to 79% fissions per initial metal atom.¹² This was initially part of the U.S. Department of Energy (DOE) Deep Burn project that undertook studies on helium-cooled pebble-bed reactors designed for deep burn.¹³ Those studies identified the challenges and the strategies to overcome those challenges.

This history combined with more recent developments of next-generation nuclear plant coated-particle fuels provides a basis to believe very high burnups are feasible. It is the only fuel with the capability for very high burnups. If it is not reprocessed, the actinide-burning SNF can be directly disposed of. As a waste form, limited studies indicate it is a superior waste form relative to other types of SNF. However, any such actinide fuel implies high power generation in the reactor with the need for high heat removal rates. The two-size pebble-bed fuel allows the normal low-enriched fuel with normal power densities to be the driver fuel and the small pebbles to be the actinide fuel with the higher power density because of their higher surface (heat transfer)-to-volume ratio. There is also the option of processing the pebble-bed fuel for recycle. There are several special considerations for this application:

1. Power density. The two pebble sizes break several traditional constraints in other actinide-burning systems. In most cases, heat-transfer constraints limit power densities in the actinide target fuel resulting in long irradiation times. The small pebbles have a much higher surface-to-volume ratio enabling better cooling of the small pebbles relative to the large pebbles, and thus higher power densities that reduce irradiation times and higher fuel loadings that reduce waste volumes. In the example above, the diameter of the small pebbles is one sixth that of the larger pebbles, with 36 times the surface area for heat removal per unit of pebble volume. However, in an accident condition, high power densities create safety challenges. The large lower-power-density-driver fuel pebbles have a higher heat capacity that provides an in-core decay heat sink to assure reactor safety for a reactor that also contains highly loaded actinide fuel. The actinide fuel could be plutonium, another actinide, or mixed actinides. A PB-HTGR or PB-FHR would allow large differences in power density between the larger pebbles that are the driver fuel and the smaller pebbles that are the actinide fuel to be burned. As a liquid-cooled reactor, the PB-FHR allows for much higher power densities than a gas-cooled reactor (3 to 10 times higher power density) and may dramatically reduce the time required to burn out actinides in a small pebble. In theory, the power density for actinide burning by combining small pebbles and salt cooling could be two orders of magnitude larger than in a conventional gas-cooled fast reactor. However, fuel limits on power density or neutronic limits on local depression of neutron flux may now control the peak power density—not the traditional heat transfer limits.

2. Uniform burnup. Pebble beds allow uniform burnup. In a typical pebble-bed reactor the pebbles are circulated through the reactor core about ten times until they reach their specified burnup. Burnup is measured by radiation detectors as each pebble leaves the reactor core. This is a capability that is required if one wants to achieve a certain fraction of actinide burnup consistently.

3. Driver fuel. In traditional reactors, the driver fuel near the target is quickly depleted. With pebble-bed fuel, the driver fuel (the large pebbles) continually moves providing fresh driver fuel for the target.

4. Spectral shift reactor. There are spectral-shift actinide burning options that vary the ratio of the numbers of the two pebble sizes, as well as including zirconium oxide or other inert coated-particle particles in the pebbles to partly replace fuel and graphite.

The separation of driver fuel and a target that can have higher power densities allows irradiation of other materials for production purposes, including production of higher actinides or other radionuclides (⁶⁰Co). The key features are (1) the decoupling of power density of the small pebbles from the large pebbles, (2) the movement of pebbles that provides fresh driver fuel on a continuous basis, and (3) separate cycling and burnup of the small pebbles. No detailed studies of these options have been completed.

III.C. Thermal-Spectrum MSR Carbon Replacement

In a MSR the graphite must be regularly replaced because of neutron damage to the graphite.¹⁴ This is a complicated operation equivalent to refueling in solid-fuel reactors. It is also a major trade-off in design. Because fuel is dissolved in the coolant without the heat transfer limits from solid fuel to coolant, higher power densities and small reactor cores are possible; but, high power densities imply high radiation levels and short graphite lifetimes. At equal power levels with a larger reactor core the power density goes down and the graphite lasts longer but there is more salt and fissile fuel in the core. A pebble-bed MSR allows on-line replacement of the pebble moderator. This enables a smaller reactor core, which in turn reduces the salt inventory and consequently the fuel inventory in the reactor.

There are constraints. With two pebble sizes the solid fractions will vary from ~60% to ~80% whereas many thermal-spectrum MSRs have graphite fractions as high as 88%. There are multiple strategies to boost the carbon-to–liquid fuel ratio. One can go to more pebble sizes for a larger graphite fraction but with the penalty of a larger pressure drop. Alternatively, the density of graphite can be increased obtaining the same neutronics with a smaller graphite volume fraction. Early nuclear graphites had densities near 1.6 g/cm³. Today nuclear-grade graphite has a density near 1.8 g/cm³. Graphite densities approaching 2.2 g/cm³ (theoretical density) are now being fabricated,¹⁵ but large-scale manufacturing viability has not been demonstrated. There has been no real incentive to develop higher-density graphites for nuclear applications.

Many thermal-spectrum MSR designs¹⁶ have a higher salt-to-graphite ratio around the outside of the core in the reflector zone. This results in a reflector zone with net absorption of neutrons. It has been proposed that this zone have pebbles and the option of replacing the moderator in this zone.

III.D. Spectral-Shift MSRs

In the last several years there has been a growing interest in MSRs where the neutron spectrum is changed over time as the fuel composition changes and fission products build up to maximize breeding and burnup for better fuel utilization.¹⁷ Spectral shifting is a partial substitute for adding fissile materials or using control rods to compensate for burnup. It widens the choice of feed materials that can be efficiently used including LWR SNF. These proposals change the neutron spectrum by withdrawal of moderator from the core, which is a mechanically complex operation. If one can vary the in-core moderator-to-fuel coolant ratio over time by changing the ratio of small-to-large carbon pebbles, a simple method for changing the neutron spectrum would exist and create a more practical method of spectral shift of the neutron flux in a MSR.

III.E. In Situ Coolant Cleanup (Tritium, Noble Metals, Other)

In helium- and salt-cooled pebble-bed reactors, tritium is produced. In both reactors tritium can leak from the fuel. In HTGRs,³He is converted into tritium. In FHRs, large quantities of tritium are produced if the coolant is a lithium or beryllium salt. In MSRs there is fission product tritium and tritium production from the coolant if a lithium or beryllium salt coolant is used. In all graphite reactors, there is significant sorption of the tritium by the graphite.

Recent experiments^18,19 in the last 2 years have shown that at 700°C some types of carbon have hydrogen sorption capabilities 50 times that of traditional nuclear-grade graphite materials (Fig. 2). This may create the option of manufacturing the small nonfuel pebbles out of carbon designed to absorb tritium and other impurities in the coolant in-core, while in the FHR and HTGR the larger pebbles contain the fuel. The small pebbles, after traveling through the reactor core, would be baked out at higher temperatures to remove the impurities before being returned to the reactor core.

Fig. 2. Hydrogen sorption at 700°C on different carbon forms. IG-110U is a traditional nuclear graphite with hydrogen sorption nearly on the x-axis.

With smaller pebbles of tritium-adsorbing carbon, the total tritium release to the power cycle may be reduced by more than an order of magnitude. Tritium removal is maximized by (1) the choice of carbon, (2) the high surface-to-volume ratio of the small pebbles that maximizes mass transfer from coolant to the pebble surface, (3) the highly turbulent flow in a pebble-bed reactor with good mass transfer of tritium from coolant to carbon, and (4) the lack of fuel in the small pebbles resulting in pebble temperatures very near coolant temperatures. Tritium sorption is lower at higher temperatures. If tritium and other impurities are captured in the core and cooled in storage, the risk of their escape is much lower than allowing them to circulate in the coolant to be eventually removed by an external-to-the-core coolant cleanup system, particularly for tritium that diffuses through metal hot heat exchangers. This option could significantly reduce secondary waste management constraints associated with these reactors.

There may be the option of incorporating some of these tritium-adsorbing carbons in the pebble-bed fuel and use of the fuel as a tritium removal system. Nuclear-grade graphite is required for reflectors and structural materials in an FHR to maximize lifetime. The irradiation limit for the pebbles is the fast neutron fluence, but the pebbles are discharged as SNF before reaching those limits so there is the potential for some relaxation of the carbon specifications for the pebbles.

In MSRs the fission process generates highly radioactive noble metal fission products that plate out everywhere in the primary system.²⁰ This results in extremely high levels of radiation associated with the heat exchangers. It also implies a potential radiation damage mechanism to heat exchangers from beta particles. The noble metals preferentially plate out on metal surfaces but there is some plate out on carbon surfaces. In the last 20 years there has been development of carbon catalysts in the chemical industry where the carbon provides the high surface area and nickel is incorporated into the surface as the catalyst. Nickel is compatible with high-temperature salts and provides a surface suitable for noble metal plate out. The small pebbles may be used as an in situ noble metal cleanup system. Noble metals generally enable faster transfer of tritium from the coolant to carbon by catalyzing H₂ dissociation and thus accelerating tritium removal, but the noble metals produce decay heat that may raise pebble temperatures.

Last, and much more speculative, is the potential to selectively remove other fission products from MSRs. This is based on two observations. First, various forms of carbon are used in industry to selectively remove impurities from gas and liquid streams including organics and inorganics. Second, recent work²¹ on understanding the mechanisms of tritium removal has begun to couple the structure of the pore structure of the carbon to its efficiency in tritium removal. This enables developing carbons for tritium removal but also raises the question of whether we can modify the carbon for removal of other classes of fission products. The understanding of carbon, the experimental methods, and the computational tools to assess such options did not exist a decade ago.

III.F. Expanded Coolant Options for FHRs

The baseline coolant in the FHR is Li₂BeF₄ (flibe) that uses isotopically separated ⁷Li to minimize neutron adsorption, which is an expensive coolant. Using two pebble sizes would reduce the coolant volume in the core from ~37% to near 20% (see Sec. II). This would have several major impacts:

1. Tritium production. The tritium generation rate is directly proportional to the coolant volume in the core. Reducing the coolant volume by almost a factor of 2 would reduce tritium production by almost a factor of 2.

2. Salt costs. Reducing salt inventories reduces salt costs. Reducing coolant inventories is important in both FHRs and MSRs. Historically MSR designs resulted in about half the fuel coolant inventory inside the core and half outside of the reactor core implying lowering coolant inventories outside of the core is as important as lowering coolant inventories inside the reactor core. Whether large reductions in out-of-core coolant inventory is viable depends upon whether compact heat exchangers can be used with their very low coolant inventories. In any case, there exist incentives to lower in-core coolant inventories.

3. Salt choices. There are alternative coolants using sodium and zirconium but these lower-cost salts are inferior to flibe.^22,23 The neutron absorption is greater and thus requires higher fuel enrichments. The designs are more tightly constrained because of the potential of a positive void coefficient in coolants with higher neutron adsorption cross sections. Reducing in-core coolant volume would significantly improve the viability of these alternative coolants by reducing parasitic neutron adsorption.

IV. CONCLUSIONS

A pebble-bed reactor with two or more sizes of pebbles creates options that have not been previously explored. In pebble-bed reactors, the greater heat transfer per unit volume of the smaller pebbles enables higher power densities relative to the large pebbles that enables higher power densities in the reactor core (smaller reactor) and creates a new pathway to actinide burning. The small pebbles could be used for coolant cleanup, from tritium to noble metals. In MSRs, multiple pebble sizes may address the challenge of replacing graphite damaged by neutron irradiation, a design option that translates into smaller reactor cores that implies less salt and a lower reactor fissile fuel inventory.

The enabling technologies are the building of experimental facilities to understand pebble-bed movement and advances in computational modeling that provide the tools to understand what is possible and what is not in terms of moving random-packed pebble beds, heat transfer, and pressure drop. Such capabilities did not exist until very recently. Research is required to determine what options are potentially feasible and whether any of the options are worthy of ultimate deployment. This must be followed by significant experimental work to confirm that pebble-bed behavior with multiple size pebbles can, in fact, be predicted under the full range of possible operating conditions.

Acknowledgments

We would like to thank the DOE Office of Nuclear Energy and the Shanghai Institute of Applied Physics (SINAP) of the Chinese Academy of Science for their support.