The reactor physics characteristics of a transuranic mixed oxide fuel in a heavy water moderated reactor

Nuclear Engineering and Design 241 (2011) 3768–3776

Contents lists available at ScienceDirect

Nuclear Engineering and Design journal homepage: www.elsevier.com/locate/nucengdes

The reactor physics characteristics of a transuranic mixed oxide fuel in a heavy water moderated reactor A.C. Morreale, W.J. Garland, D.R. Novog ∗ Department of Engineering Physics, McMaster University, 1280 Main St. W. Hamilton, Ontario, Canada L8S 4L7

a r t i c l e

i n f o

Article history: Received 24 January 2011 Received in revised form 7 July 2011 Accepted 8 July 2011

a b s t r a c t The reprocessing actinide materials extracted from spent fuel for use in mixed oxide fuels is a key component in maximizing the spent fuel repository utility. While fast spectrum reactor technologies are being considered in order to close the fuel cycle, and transmute these actinides, there is potential to utilize existing pressurized heavy water reactors such as the CANDU® 1 design to meet these goals. The use of current thermal reactors as an intermediary step which can burn actinide based fuels can significantly reduce the fast reactor infrastructure needed. This paper examines the features of actinide mixed oxide fuel, TRUMOX, in a typical CANDU nuclear reactor. The actinide concentrations used were based on extraction from 30 year cooled spent fuel and mixed with natural uranium in 4.75% actinide MOX fuel. The WIMS-AECL model of the fuel lattice was created and the two neutron group properties were transferred to RFSP in order to create a 3 dimensional time average full core model. The model was created with typical CANDU limits on bundle and channel powers and a burnup target of 45 MWd/kgHE. The TRUMOX fuel design achieved its goals and performed well under normal operations simulations. This effort demonstrated the feasibility of using the current fleet of CANDU reactors as an intermediary step in burning reprocessed spent fuel and reducing actinide burdens within the end repository. The recycling, reprocessing and reuse of spent fuels produces a much more sustainable and efficient nuclear fuel cycle using existing and proven reactor technologies. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Recently, drivers such as climate change and the desire for secure and stable energy production have generated increased interest in nuclear energy. The issues of spent fuel and the sustainability of the once through fuel cycle utilized in current nuclear power facilities need to be addressed. Fuel used in a once through cycle still contains a significant portion of its initial energy potential, as well as potential from transuranic actinides that are produced during the fission process. Hence it can be used in recycle and reuse programs to improve fuel utilization and reduce spent fuel repository thermal and radiological burdens per MWe produced. Reprocessing is a vital step towards closing the nuclear fuel cycle and reducing the volume of end waste. The actinide materials can be extracted from current inventories of LWR spent fuel and placed in a mixed oxide matrix and utilized in the current fleet of pressurized heavy water reactors.

∗ Corresponding author. Tel.: +1 905 525 9140. E-mail addresses: [email protected] (A.C. Morreale), [email protected] (W.J. Garland), [email protected] (D.R. Novog). 1 CANDU is a trademark of Atomic Energy of Canada Limited. 0029-5493/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.nucengdes.2011.07.017

The reprocessing of actinides is central to the long term sustainability of nuclear power. Stockpiles of spent nuclear fuel that have cooled over many years are an excellent target for reprocessing. The materials can be separated into streams for recycle, reuse and waste. In particular, Uranium and useful actinides such as Americium, Curium, Neptunium and Plutonium can be extracted from the spent fuel and reused. The total end waste is the remaining fission products present in the fuel which accounts for ∼3–5% of the original mass, depending on initial enrichment and burnup lifetime (Wilson, 1996). The actinides have long lasting radioactive loads and are primary contributors to the spent fuel heat load. Therefore, reprocessing significantly reduces the amount of waste, activity level and heat load demands on long term spent fuel storage facilities.

1.1. Strategies of actinide reprocessing The governing factors of reprocessing include the type of recycled fuel matrix, reactor type and the choice of sustained, continuous or limited recycle iterations. The actinide materials can be blended into a fuel matrix of an inert material such as Zirconium or Silicon (IMF) or a Uranium mixed oxide (MOX). The non-fertile IMF does not generate more actinides during burnup but

A.C. Morreale et al. / Nuclear Engineering and Design 241 (2011) 3768–3776

presents different thermal characteristics and neutronic properties which complicate its behaviour in-core. The MOX form blends the actinides with standard UO2 fuel producing a fuel that retains the overall thermal and neutronic properties of standard fuel. The MOX format has been extensively studied and utilized in weapons reprocessing efforts and was found to behave similar to standard fuel. The fuel mixture can be utilized in homogenous elements with all the pins the same or heterogeneous configurations. The recycle processes differ in cycle length and the extent of usage of reprocessed waste. A limited cycle extracts usable actinides from cooled spent fuel and integrates it with newly mined ore to be run through reactors. The fuel is recycled for a few iterations and then the end stage spent fuel is sent to a long term depository. Continuous strategies follow the same steps but continue to extract the actinides and only send the waste fission products to depositories (i.e. reusing the remaining fissile content). Sustained recycle methods build on this by using fast breeders which produce new fissile material as part of continuous operations. The minor actinide concentration increases with each recycling operation, affecting the fuel neutronics and complicating the spent fuel separation efforts. Thermal plants are plentiful and readily available worldwide but are unable to cope with more than a few iterations of actinide recycle before safe operation concerns arise (Weigland et al., 2007). Fast breeder designs are more complex and rare but are able to perform repeated and continuous recycle. There is potential for synergy between the two reactor types by using thermal reactors in the first couple of iterations to reduce the volume of waste as well as the amount of fast reactor infrastructure required. 1.2. Benefits and challenges of reprocessing The reprocessing of spent fuel and movement away from a once through fuel cycle provides benefits related to reduction of overall waste and the more efficient utilization of the energy in fuel. The current once through nuclear generation cycle in the United States outputs about 100 GWe resulting in 2200 tons of heavy metal (THM) of spent fuel per year (Baetsle and De Raedt, 1997). The proposed long term repositories have specific storage and temperature limits, which in turn dictate the load of spent fuel that can be accommodated. The Yucca Mountain project had an administrative maximum capacity of 70,000 THM and a temperature limit between adjacent waste rows of 96 ◦ C, the local boiling point of water (Weigland et al., 2007; Baetsle and De Raedt, 1997). Using these limits as a guideline, and assuming the current rate of nuclear power production remains constant, a repository of the size of Yucca Mountain would need to be opened every 32 years. Considering the swell of interest in nuclear energy as a low carbon emission power production strategy, this is likely an underestimate of the repository needs. Additionally, in the once through cycle there is still significant energy retained in the spent fuel, making it a potential future resource. Actinide extraction, along with other recycle operations, can reduce the total end waste to only the fission products which make up a small portion (3–5 wt%) of spent fuel and a smaller fraction of the heat load. Proper use of this strategy could increase the repository loading by a factor of 200 (Weigland et al., 2007), reducing the demand for large repository facilities and avoiding disposing of useful material. These estimates rely on the use of a continuous recycle strategy to attain maximum benefit. A limited recycle strategy only reduces the repository demands by half due to the fact that actinide based spent fuel requires 10–20 times more repository space due to increased heat load (Weigland et al., 2007). The other benefit of a closed cycle is the fact that current inventories of spent fuel become an important commodity which supplements freshly mined sources; thus allowing the nuclear industry to cope

3769

better with increased demand. In general, actinide fuels tend to have a higher fissile content allowing for increased burnup levels and more efficient energy utilization. Reprocessing despite its efficiency and economic benefits presents unique challenges. The partitioning and transmutation activities required to extract the actinides from spent fuel are complex. There is also extensive chemical handling involved and at all times criticality and radiation protection concerns must be addressed combining the hazards of large scale chemical processing and nuclear fuel production as well as the associated waste streams generated. Current methods require stringent controls and have high costs which are important issues in determining the viability of reprocessing. Specific reprocessing methods are not explored here as the focus of this work is on the feasibility of utilizing reprocessed spent fuel not its fabrication. 1.3. Reactor technology for actinide burning The use of thermal water moderated reactors as a first stage in improved utilization allows 1–2 iterations of reprocessed spent fuel to be burned before requiring fast breeder reactors to close the cycle. This significantly reduces the fast breeder infrastructure needed, if used for the first cycle alone the actinide inventory in spent fuel is reduced as the production of one actinide MOX bundle requires the actinide content of 13.5 regular spent fuel bundles (Weigland et al., 2007). The remaining materials in these bundles can be recycled (e.g. uranium) while the waste fission products are sent to long term disposal. In addition, a significant advantage is that these thermal reactors are a proven and safe technology currently available and can be utilized with a full core of MOX bundles or with only a portion of the fuel being MOX without major changes to design or operations. Efforts of reprocessing nuclear weapons material and early trials of recycled spent fuel have been employed in current reactors with 5–10% of the fuel in the core being MOX. Comparisons of plutonium burning in heavy and light water reactor designs have been explored previously (Taiwo et al., 2007) and both technologies show promise in the area of initial actinide burning. Therefore, the logical first step of thermal reactor utilization can be performed using the MOX designs currently employed in pressurized water reactors (PWR), boiling water reactors (BWR) or pressurized heavy water reactor (PHWR) designs such as CANDU. The CANDU design offers extensive flexibility due to high neutron economy, a simple compact fuel design and the availability of online fueling. The neutron economy produced by the combination of heavy water coolant and moderator along with low-neutron absorbing materials provides for efficient harnessing of the fuel at various levels of enrichment. Online fueling reduces the need for poison loading (i.e. Boron in the moderator or coolant water) to counteract fresh fuel reactivity and provides constant management of the flux profile in the core. The fuel design of 37 or 43 elements in a circular multi-ringed arrangement is simple, allows for the inclusion of integrated poison and various fuel contents and enrichments. Bundles can be optimized for linear power rating, coolant void reactivity or differential burnup. The adaptability and flexibility of the design make CANDU an excellent choice for the employment of reprocessed actinide fuel. 2. Feasibility of actinide fuel in CANDU 2.1. Design description This feasibility study examines the use of actinide MOX fuel in the standard CANDU pressurized heavy water reactor. Americium, Curium, Neptunium and Plutonium are extracted from 30 year cooled spent LWR fuel and blended with natural Uranium to

3770

A.C. Morreale et al. / Nuclear Engineering and Design 241 (2011) 3768–3776

Fig. 1. Flow chart of TRUMOX design and analysis.

produce the actinide based MOX fuel, referred to as transuranic mixed oxide fuel or TRUMOX. The fuel is designed for high burnup and has a target of 45 MWd/kgHE which is 6 times the level of standard natural uranium CANDU fuel. In order to evaluate a fuel design there is a three phase procedure to be followed. The first is fuel composition and bundle design to meet specific burnup, fissile content and reactivity properties and is performed using a lattice cell code. The second phase uses a diffusion code to examine implementation of the fuel design into a CANDU core that meets current regulatory limits. The third phase evaluates the operational behaviour and physics characteristics of the design including reactor coefficients, control system responses and fueling. This paper modifies an existing lattice model for the first phase and primarily focuses the second and third phases. A flow chart of the design and analysis is provided in Fig. 1. The design utilized is a 43 element bundle with mixed actinide fuel in all elements except the center pin which contains a burnable poison that is used to reduce coolant void reactivity (CVR) effects. The lattice cell calculations are performed using WIMS-AECL v3.1. WIMS-AECL is a 2D multi group transport code and, for this study, was used with the ENDFB-VI library modified with patches for Dysprosium and Curium. A standard CANDU design (i.e. 6 m diameter Calandria, 6 m long fuel channels, 21 adjusters and 380 fuel channels) was used in the study and modeled using the RFSP-IST 2 group neutron diffusion code. Both WIMS-AECL and RFSP are part of the industry standard tool set for reactor analysis in Canada. The design limits include a burnup target of 45 MWd/kgHE, a mean CVR target of 5 mk and power limits of 935 kW/bundle and 7.3 MW/channel (the administrative limits for CANDU). The high burnup target of this study may not be compatible with the existing CANDU fuel carrier design and material performance. Current CANDU reactor fuel designs run to burnup levels of ∼10 MWd/kgHE and hence may not be suitable at the higher level of burnup used in this study. The TRUMOX fuel design assumes that a suitable fuel carrier material and design will be available that can withstand these high burnup levels perhaps utilizing some of the features of PWR fuel carrier materials and designs which regularly reach the level of burnup proposed in this study. 2.2. WIMS-AECL fuel design The fuel bundle is based on the 43 element CANFLEX design with Dysprosium burnable poison in the center element. The center pin is thicker with an outer diameter of approximately 17.4 mm while the other pins are approximately 11.4 mm OD and are arranged in

Fig. 2. TRUMOX 43 element bundle design.

three rings of 7, 14 and 21 elements around the central pin. Fig. 2 provides a depiction of the bundle arrangement. The WIMS model used 89 energy groups with a meshing of 53 lines and 11 angles within the calandria tube. The numerical accuracy of this meshing was explored and for meshing from 50 to 5000 lines and 11, 13, 23, and 31 angles, the K-infinity value stays within 0.65 mk of the original mesh which is accurate enough for this feasibility study. The actinide composition is based on data from Oak Ridge National Laboratories that predicts the probable yields of actinides from spent fuel reprocessing (Collins et al., 2004). The compositions of the fuel calculated in weight percent is provided in Table 1. The final TRUMOX fuel design used was 95.25 wt% UO2 and 4.75 wt% AOX containing 88.153 wt% heavy elements and 11.847 wt% O2 with a total mass of heavy metals per bundle of 17.362 kg. In order to meet the high burnup target the fissile elements in this fuel (Pu239 , Pu241 , and U235 ) make up 3.49% of the heavy elements compared to only 0.71% in natural uranium CANDU fuel. The average full core exit burnup achieved in the design was 43.429 MWd/kgHE (96.5% of the target). Details of the WIMS-AECL model are provided in Table 2 and the level of actinide burnup is provided in Table 3. 2.3. RFSP-IST full core modeling The full core modeling was performed with the reactor fueling simulation program, industry standard toolset (RFSP-IST). RFSP is a 2 neutron energy group diffusion code that is capable of modeling neutron spatial and temporal behaviour. The main inputs to RFSP include core geometry, fuel and moderator properties and fuel burnup effects from WIMS, online fuelling patterns and irradiation targets as well as incremental cross sections for reactivity control devices. The CANDU model contained the in-core structures, reactor control and safety systems as well as the standard fuel channel and calandria geometry. The study was restricted to a typical CANDU with no physical changes to the core structures (i.e. no reactor design modifications were considered). External to the RFSP-IST code, a set of super cell calculations is usually performed by DRAGON to produce the differential cross-sections produced by the interaction of control devices with the lattice cells. The supercell calculations were not recomputed for the TRUMOX fuel design and the standard natural uranium fuel values were used. This was

A.C. Morreale et al. / Nuclear Engineering and Design 241 (2011) 3768–3776

3771

Table 1 Fuel composition, actinide uranium and dysprosium zirconium oxide. Actinides

Uranium mix

Dysprosium zirconium oxide

Isotope

Type

wt%

Isotope

Type

wt%

Isotope

Type

wt%

Np-237 Pu-238 Pu-239 Pu-240 Pu-241 Pu-242 Am-241 Am-243 Cm-243 Cm-244 Cm-245 Cm-246

Actinide Actinide Actinide Actinide Actinide Actinide Actinide Actinide Actinide Actinide Actinide Actinide

4.698 1.301 56.243 20.099 3.040 3.800 9.907 0.763 0.001 0.072 0.012 0.001

U-234 U-235 U-238

U Mix U Mix U Mix

0.0054 0.7110 99.2836

Dy-160 Dy-161 Dy-162 Dy-163 Dy-164 Zr-90 Zr-91 Zr-92 Zr-94 Zr-96 O-16

Absorber Absorber Absorber Absorber Absorber Zr Mix Zr Mix Zr Mix Zr Mix Zr Mix Oxygen

1.346 10.943 14.858 14.597 16.633 12.388 2.732 4.221 4.371 0.719 17.192

Actinides Oxygen

Actinide Oxide (AOX) 88.207 11.793

wt% wt%

Uranium Oxygen

Uranium Oxide (UO2 ) 88.150 11.850

wt% wt%

Table 2 WIMS-AECL model lattice cell characteristics. Parameter

Approximate Value

Coolant density Coolant temperature Coolant purity Moderator density (variable) Moderator temperature Moderator purity Fuel temperature Fuel density (4.75% actinide) Volume of fuel per bundle

0.807 g/cm3 288.1 ◦ C 98.8% D2 O 1.089 g/cm3 (avg) 63 ◦ C 99.92% D2 O 587 ◦ C 10.54 g/cm3 1869.44 cm3

deemed unnecessary for the feasibility study as it focuses on the core predictions with as few internal control devices in the core as possible to allow maximal burnup. Ongoing work will address the changes in the super-cell results. In order to accommodate the TRUMOX fuel, fuelling patterns, target exit irradiations for fueling and control device changes were considered. The fuel design contains an integral burnable poison in the central pin which produces a flat flux profile across the bundle and in general provides a flatter axial profile along each channel. Therefore, the adjuster rods used to flatten the flux for standard natural uranium fuel were not needed and the models were run with the adjusters out of the core. In the interest of easy transition to TRUMOX, efforts were made to maintain standard operating characteristics of a natural uranium fuelled CANDU as much as possible. Identical geometry with respect to fuel channel and control structures was important in this feasibility study, subsequent studies will investigate the impact of changing control device geometry (i.e. adjuster rods).

2.3.1. Model design The model portion of RFSP-IST allows the partitioning of the core into regions and to define the fueling scheme for these regions. The TRUMOX fuel is much more active than standard fuel (due to its higher enrichment) and hence requires a modified fueling scheme with 1–2 bundle shifts over 15 defined regions, compared to the 7 region, 4–8 bundle shift strategy normally used with natural uranium (NU) fuel. For the TRUMOX design, nine of the regions utilize one-bundle shifts and the remaining six use two-bundle shifts resulting in 264 one-bundle-shift channels and 116 two-bundleshift channels. The region and fueling information for the TRUMOX core is provided in Fig. 3. 2.3.2. Time average results The time average module determines the total core power distribution in the radial and axial directions and can be used to refine the fuelling and burnup patterns such that the bundle and channel power limits of the TRUMOX fuel mimic those of standard NU fuel. The options include the control of long term reactivity control device positions such as the adjuster rods, and the liquid control zone levels. The time average module was finalized with all the adjuster rod banks removed from the core and the final average burnup reaching the aforementioned value of 43.4 MWd/kgHE. The removal of the adjusters left the liquid zone controllers as the only in core reactivity devices considered. This allowed the core to achieve the maximum burnup possible. The integral burnable neutron poison in the fuel provided flux flattening, allowing the adjusters to be left out of the core. The irradiation characteristics for each region are provided in Table 4 and the 3-D channel power profile for the TRUMOX core is shown in Fig. 4. A plot of the axial

Table 3 Actinide burnup for 43.4 MWd/kgHE. Concentration

U 235

Np Tot

Pu Tot

Am Tot

Cm Tot

Initial Final Change (final–initial) % (change/initial)

2.38E+00 4.33E−01 −1.95E+00 −81.79%

7.56E−01 3.92E−01 −3.64E−01 −48.14%

1.36E+01 7.96E+00 −5.63E+00 −41.44%

1.72E+00 6.49E−01 −1.07E+00 −62.22%

1.39E−02 3.43E−01 3.29E−01 2374.41%

Table 4 TRUMOX CANDU time-average exit irradiation values. Region

Average exit irradiation (n/kb)

Region

Average exit irradiation (n/kb)

Region

Average exit irradiation (n/kb)

TOP EXT TOP MID TOP TOP OUT OUTER

3.643 3.672 3.804 3.834 3.883

OUTER RING MIDDLE MID OUT INNER HI

3.643 4.767 4.72 4.43 4.266

RIGHT BOTTOM BOTTOM OUT BOTTOM MID BOTTOM EXT

3.637 3.819 3.812 3.439 3.114

3772

A.C. Morreale et al. / Nuclear Engineering and Design 241 (2011) 3768–3776 FUELLING SCHEME CORE MAP- 15 REGIONS, 1 & 2 BUNDLE SHIFTS 2

1

3

4

5

6

7

8

9

10 2BS

A 2BS

B

12

13

TOP-EXT

11

2BS

17

18

19

21

22 A B C

TOP 1BS

D

TOP OUT 1BS

HI

1BS

E 2BS

2BS

F

20

1BS

TOP OUT 1BS

16

2BS

1BS

E

15

TOP-MID

C D

14

1BS

G

F

1BS

MIDDLE

G H

H 2BS

J

1BS

1BS

1BS

1BS

K

1BS

L

OUTER

M

RING

OUTER

HI

1BS

2BS

1BS

MID OUT

J K

INNER

MID OUT

HI

OUTER

RIGHT

L M

1BS

N 2BS

O

1BS

1BS

1BS

1BS

N

1BS

1BS

1BS

1BS

2BS

O P

P 1BS

Q

1BS

Q

2BS

R

2BS 1BS

S T

1BS

BOTTOM OUT

1BS

HI BOTTOM

1BS

U 2BS

V

2BS 2

1

3

4

5

6

7

8

9

1BS

U

11

12

V

2BS

W

2BS

BOTTOM-EXT 10

S T

BOTTOM OUT

BOTTOM-MID

W

R

1BS

13

14

15

16

17

18

19

20

21

22

Fig. 3. TRUMOX CANDU, fuel irradiation regions and associated fueling scheme.

thermal flux is provided in Fig. 5 and the characteristics of the time average core are provided in Table 5. The time average results provide the mean channel power behaviour but do not include local spikes which may occur due to periodic on-line refueling. As a result, channel and bundle powers must be investigated at random times during potential fuelling operations to ensure the instantaneous power profiles conform to existing license constraints. This is defined as the ripple effect and is analyzed using the INSTANTAN module of the RFSP code, which predicts the core behaviour accounting for online refueling effects. 2.3.3. RFSP-IST INSTANTAN simulations The INSTANTAN module allows the production of a random snapshot of the core at a given time with some channels recently fuelled, some in the middle of the fueling cycle and some high bur-

nup channels that are to be fuelled soon. A random aging pattern is provided which dictates the stage in the fueling cycle for each channel. This type of analysis simulates a random core configuration in time and ensures that the fuelling and irradiation parameters are able to maintain the criticality of the core. Furthermore, these simulations provide data to ensure the bundle and channel power

6500 6000 5500 4500 4000 3500 3000 2500 2000 1500 1000 500 0

A E J N Channel Row

R V 1

3

5

7

9

11

13

15

17

19

Channel Power (kW)

5000

Fig. 5. TRUMOX CANDU, time average axial thermal flux in the vertical midplane.

6000-6500 5500-6000 5000-5500 4500-5000 4000-4500 3500-4000 3000-3500 2500-3000 2000-2500 1500-2000 1000-1500 500-1000 0-500

21

Channel Column

Table 5 TRUMOX CANDU time-average characteristics. Characteristic

TRUMOX value

Std CANDU 6 value (Rouben, 2002)

K-effective Max BP Max CP Avg whole core exit burnup Maximum exit burnup Radial Form Factora Axial Form Factorb Overall Form Factorc

0.999478 818 kW 6298 kW 43.43 MWd/kgHE 47.44 MWd/kgHE 0.861 0.642 0.553

∼1 800–820 kW 6500–6600 kW 7.5 MWd/kgHE ∼8.00 MWd/kgHE 0.824 0.684 0.564

a b

Fig. 4. TRUMOX CANDU, time-average core channel powers, 100%FP.

c

Radial Form Factor = Average Channel Power/Maximum Channel Power. Axial Form Factor = Max Channel Power/(Max Bundle Power × # of bundles). Overall Form Factor = Average Bundle Power/Maximum Bundle Power.

A.C. Morreale et al. / Nuclear Engineering and Design 241 (2011) 3768–3776

3773

Table 6 Sample PRCAD matrix with element (1) as the oldest channel.

M N O P Q R S

A B C D E F G H J K L M N O P Q R S T U V W

1

5

6

7

8

9

10

11

(1) 0.98 (8) 0.60 (15) 0.34 (22) 0.72 (29) 0.58 (36) 0.20 (43) 0.74

(2) 0.24 (9) 0.46 (16) 0.88 (23) 0.06 (30) 0.28 (37) 0.94 (44) 0.40

(3) 0.66 (10) 0.04 (17) 0.14 (24) 0.44 (31) 0.82 (38) 0.12 (45) 0.62

(4) 0.84 (11) 0.30 (18) 0.76 (25) 0.54 (32) 0.16 (39) 0.70 (46) 0.36

(5) 0.38 (12) 0.56 (19) 0.22 (26) 0.96 (33) 0.42 (40) 0.08 (47) 0.90

(6) 0.92 (13) 0.10 (20) 0.68 (27) 0.02 (34) 0.64 (41) 0.18 (48) 0.52

(7) 0.50 (14) 0.80 (21) 0.26 (28) 0.48 (35) 0.86 (42) 0.32 (49) 0.78

0 0 0 0 0 0 0 0 0.16 0.70 0.36 0.72 0.06 0.44 0 0 0 0 0 0 0 0

2

0 0 0 0 0 0 0.22 0.96 0.42 0.08 0.90 0.58 0.28 0.82 0.16 0.42 0 0 0 0 0 0

3

0 0 0 0 0.92 0.10 0.68 0.02 0.64 0.18 0.52 0.20 0.94 0.12 0.70 0.08 0.18 0.32 0 0 0 0

4

0 0 0 0.78 0.50 0.80 0.26 0.48 0.86 0.32 0.78 0.74 0.40 0.62 0.36 0.90 0.52 0.78 0.50 0 0 0

5

0 0 0.20 0.74 0.98 0.24 0.66 0.84 0.38 0.92 0.50 0.98 0.60 0.34 0.72 0.58 0.20 0.74 0.98 0.24 0 0

6 0 0.28 0.94 0.40 0.60 0.46 0.04 0.30 0.56 0.10 0.80 0.24 0.46 0.88 0.06 0.28 0.94 0.40 0.60 0.46 0.04 0

7 0 0.82 0.12 0.62 0.34 0.88 0.14 0.76 0.22 0.68 0.26 0.66 0.04 0.14 0.44 0.82 0.12 0.62 0.34 0.88 0.14 0

8 0 0.16 0.70 0.36 0.72 0.06 0.44 0.54 0.96 0.02 0.48 0.84 0.30 0.76 0.54 0.16 0.70 0.36 0.72 0.06 0.44 0

9 0.96 0.42 0.08 0.90 0.58 0.28 0.82 0.16 0.42 0.64 0.86 0.38 0.56 0.22 0.96 0.42 0.08 0.90 0.58 0.28 0.82 0.16

10 0.02 0.64 0.18 0.52 0.20 0.94 0.12 0.70 0.08 0.18 0.32 0.92 0.10 0.68 0.02 0.64 0.18 0.52 0.20 0.94 0.12 0.70

11 0.48 0.86 0.32 0.78 0.74 0.40 0.62 0.36 0.90 0.52 0.78 0.50 0.80 0.26 0.48 0.86 0.32 0.78 0.74 0.40 0.62 0.36

12 0.84 0.38 0.92 0.50 0.98 0.60 0.34 0.72 0.58 0.20 0.74 0.98 0.24 0.66 0.84 0.38 0.92 0.50 0.98 0.60 0.34 0.72

13 0.30 0.56 0.10 0.80 0.24 0.46 0.88 0.06 0.28 0.94 0.40 0.60 0.46 0.04 0.30 0.56 0.10 0.80 0.24 0.46 0.88 0.06

14 0.76 0.22 0.68 0.26 0.66 0.04 0.14 0.44 0.82 0.12 0.62 0.34 0.88 0.14 0.76 0.22 0.68 0.26 0.66 0.04 0.14 0.44

15

0 0.96 0.02 0.48 0.84 0.30 0.76 0.54 0.16 0.70 0.36 0.72 0.06 0.44 0.54 0.96 0.02 0.48 0.84 0.30 0.76 0

16

0 0.42 0.64 0.86 0.38 0.56 0.22 0.96 0.42 0.08 0.90 0.58 0.28 0.82 0.16 0.42 0.64 0.86 0.38 0.56 0.22 0

17

0 0.08 0.18 0.32 0.92 0.10 0.68 0.02 0.64 0.18 0.52 0.20 0.94 0.12 0.70 0.08 0.18 0.32 0.92 0.10 0.68 0

18

0 0 0.52 0.78 0.50 0.80 0.26 0.48 0.86 0.32 0.78 0.74 0.40 0.62 0.36 0.90 0.52 0.78 0.50 0.80 0 0

19

0 0 0 0.74 0.98 0.24 0.66 0.84 0.38 0.92 0.50 0.98 0.60 0.34 0.72 0.58 0.20 0.74 0.98 0 0 0

20

0 0 0 0 0.60 0.46 0.04 0.30 0.56 0.10 0.80 0.24 0.46 0.88 0.06 0.28 0.94 0.40 0 0 0 0

21

0 0 0 0 0 0 0.14 0.76 0.22 0.68 0.26 0.66 0.04 0.14 0.44 0.82 0 0 0 0 0 0

22

0 0 0 0 0 0 0 0 0.96 0.02 0.48 0.84 0.30 0.76 0 0 0 0 0 0 0 0

Fig. 6. Sample channel age map for INSTANTAN simulation.

limits are met for a given set of random channel ages. To realistically simulate random ages for the channels that would arise after a long period of online refueling, the following procedure is applied. A 7 × 7 matrix of random ages (between 0 and 1) is generated and referred to as a “pattern random channel distribution” (PRCAD) matrix. One of the 49 (7 × 7 = 49) elements in the matrix is assigned the highest number and is hence next to be fueled. The values within this “patterned” matrix are not just random numbers they are based on an operating history to realistically represent the ages of channels near each other (Xing Guan et al., 1999). In the absence of a fuel history one could use a uniform random distribution over the interval (0, 1) with a different value for each channel. However, this method will frequently produce exaggerated maximum bundle and channel powers compared to a more accurate core history. This is mainly due to the fact that the random number generator may place several channels of the same age together, while a fuelling engineer avoids fuelling adjacent channels at the same time especially for higher power regions of the core. Therefore, the 49 elements of the PRCAD matrix utilized in this study were based off of a real fuelling history for the group of channels to provide a more realistic pattern of channel aging and avoid unrealistically excessive bundle and channel powers. The PRCAD matrix utilized is shown in Table 6. The 7 × 7 pattern is repeated across a 22 × 22 array that encompasses all of the channels in the CANDU core. Adjacent matrices are the transpose of those next to them. A sample channel age map is provided in Fig. 6 showing the PRCAD matrix from Table 6 applied over the core and would be considered one “snapshot” of the channel burnup profile at an instant in time. Using multiple PRCAD matrices, several random channel age patterns were produced which generated multiple core snapshots. These snapshots were used to evaluate and refine the irradiation times, and fuelling patterns as well as to assess the bundle and channel power limits. The result of these multiple random core snapshots obtained from RFSP are that an adequate K-effective value is maintained over these simulations and that the maximum

bundle and channel powers are kept under the limits for typical existing CANDU. At this stage the pre-conceptual design and assessment was complete and all the specifications and constraints were satisfied to a reasonable degree. The next step in the feasibility study evaluates the reactor control characteristics and its behaviour during normal operation situations. 3. Evaluation of preliminary TRUMOX CANDU design 3.1. Reactivity control device worth and reactor coefficients The worth of control devices are investigated using the time average model, the adjuster rods, mechanical absorbers (MCA), liquid zone controllers (LZCR) and shutdown system 1 (SDS1) were evaluated.2 The fine control of the reactor is accomplished with the LZCRs which consist of 14 vessels of H2 O distributed throughout the core. The water levels in the vessels are altered to control neutron absorption in the core and the distributed nature allows for both bulk and spatial control. The adjuster rods are 21 neutron absorbing rods arranged in 7 banks and are usually left inserted in the core to produce a flat flux profile and to provide a positive reactivity reserve to counteract Xenon buildup. These are coarse control devices and are meant to supplement the LZCRs. The MCAs are 4 neutron absorbing rods normally positioned outside the core and driven in to provide negative reactivity as a coarse control supplement to the LZCRs. SDS 1 is a set of 28 neutron absorbing rods deployed in 4 banks and are poised above the core ready to be released for shutdown purposes. The only change made to normal operating procedures with respect to the control systems is the removal of the adjusters from the core. The TRUMOX fuel includes an integrated burnable poison

2 Differential cross-sections were not updated for the TRUMOX fuel, standard NU fuel values were used. Subsequent work will re-examine the impact of the TRUMOX fuel on the neutron spectrum and control system device reactivity worth.

3774

A.C. Morreale et al. / Nuclear Engineering and Design 241 (2011) 3768–3776

Table 7 Approximate reactivity worth of control devices. Device LZCRs 0–100% Full 0–50% Full 50–100% Full Adjuster rods Mechanical absorbers (MCAs) Shutdown system (SDS#1)

TRUMOX worth (mk)

NU fuel worth (mk)a (Rouben, 2002)

3.4 2.0 1.3 2.5 3.2 34.8

∼7 n/a n/a ∼15.0 ∼10.0 ∼80.0

Note: Exit burnup values are: TRUMOX = 43.4 MWd/kgHE, NU fuel = 7.5 MWd/kgHE. a Approximate worth, slight variations between different reactors, n/a (not available).

Table 8 Reactor coefficients. Reactor coefficient Coolant Temperature Coefficient Fuel Temperature Coefficienta Moderator Temperature Coefficientb Moderator Purity Coefficient

TRUMOX value 0.03 mk/◦ C

Std value (AECL, 1995; Rouben, 2002) ∼0.018 mk/◦ C

−0.003 mk/◦ C

∼−0.015 mk/◦ C

−0.0218 mk/◦ C

∼−0.070 mk/◦ C

1.0 mk/atm%

∼34 mk/atm%

N. B. All temperature coefficients include the related density changes. a For small perturbations (±5 ◦ C), for the range 500–950 = −0.00281 mk/◦ C. b For small perturbations (±5 ◦ C), for the range 50–120 ◦ C = −0.0392 mk/◦ C.

which along with the fuel composition produces a flat flux profile that peaks much closer to the edges of the core than regular fuel. Therefore, the adjusters are not needed to flatten flux and if held in would dampen the central flux in the TRUMOX core pushing it further out to the edges. On a device by device basis, reactor control and safety rods were manipulated in the time average core to determine the difference in core reactivity caused by each perturbation. The results for the TRUMOX design are presented in Table 7 along with the standard values for natural uranium fuel. The reactivity worth of all the control devices were shown to be considerably lower than the typical values for CANDU with natural uranium fuel. The primary reason for this is the harder neutron spectrum produced by the actinide fuel which has a higher fissile content and larger amounts of Plutonium-239. The harder spectrum drives up the average thermal neutron energy (i.e. neutron velocity, v) reducing the absorption in the control devices since at thermal energies the absorption cross-sections of the control device follow a 1/v relationship. However, since the incremental cross-sections were not specifically recalculated for the TRUMOX fuel the values provided here are only approximate numbers. In addition to the device worth assessment, the important reactivity coefficients of the TRUMOX CANDU were analyzed including the coolant temperature, fuel temperature, moderator temperature and the moderator purity. The coolant void reactivity (CVR) of the fuel is also analyzed by reducing the coolant density by a factor of 1000 to simulate a bounding estimate of channel voiding. The full core CVR of the TRUMOX design was found to be 4.3 mk compared to the full core CVR for natural uranium fuel of approximately 10–15 mk (Rouben, 2002). The lower CVR is mainly the result of the integrated burnable poison in the center pin of the TRUMOX fuel bundle. This represents a potential improvement in loss of coolant accident (LOCA) response since lower CVR will lead to lower predicted fuel centerline and sheath temperatures. However, no accident analysis was performed as part of this feasibility study. The reactor coefficient values are provided in Table 8 along

with the standard natural uranium fuel values. The coefficient values computed are much different from the typical natural uranium results due to the much higher fissile content in the TRUMOX fuel, the harder neutron spectrum and the increased amount of Pu-239 present. The fuel temperature coefficient is only 20% of the typical value since Pu-239 which has a smaller fuel temperature reactivity response is the primary fissile element in the fuel. Moderator purity has a much lower effect with the TRUMOX fuel due to the higher fissile content and greater fuel reactivity. 3.2. Analysis of normal operating conditions As discussed previously, CANDU type reactor designs fuel online which may lead to local power peaking during these fuelling operations. These fluctuations during fuelling are referred to as the “fuelling ripple” factors in operating reactors and typically vary between 0.9 and 1.1. Hence it is necessary to study the effects of fueling the core on the bundle and channel powers and the control systems to ensure the reactor remains within operational limits with the TRUMOX fuel. Online fueling allows the addition of fuel as it is burnt up rather than the batch fueling process common in LWR designs which uses burnable neutron poisons to hold down the reactivity of fresh fuel and burn off as the fuel burns up and becomes less reactive. The standard CANDU with natural uranium fuel has a fuelling rate of about 15 bundles/full power day (FPD), fueling 2–3 channels with 4–8 bundle shifts (Rouben, 2002). The TRUMOX fuel is much more reactive and requires less fueling but since it is accomplished by 1 and 2 bundle shifts the fueling rate, in terms of channels visited per day, is slightly higher. TRUMOX has a reactivity decay rate of 0.275 mk/full power day (FPD) resulting in a recommended fuelling rate of 2.5–3 bundles/FPD. This means that given a full week of operations 17–18 channels are fueled with 20 bundles. The response to day to day fueling of channels was studied to test the short-term response of the core, fuel and LZCRs using a detailed two day fuelling simulation. This evaluation is different from the INSTANTAN studies as it looks at the direct short term control responses of the liquid zones to a fuelling event in the core. While the initial core used is one of the INSTANTAN snapshots, a time dependent RFSP simulation is performed wherein fuel is inserted into a channel and the direct change on channel power is observed. The key objective of these fuelling studies is to ensure that the control system is able to compensate for the fuelling operations as well as to ensure that the channel and bundle power limits are maintained. Sample fuelling operations were conducted for two days using 16 consecutive simulations determining the core response in detail throughout the two days. The first day, involved three 1-bundle fuelling shifts in channels S05, L12 and E11. The fuelling began at 9:00 am and each shift was spaced 3 h apart and then the core was run normally until the end of the day. The second day involved the fuelling of 2 channels with 3 bundles beginning at 9:00 am with a 1-bundle shift into channel M04 followed 3 h later with channel D05 being fuelled with a 2-bundle shift. Throughout the 2-day simulation the LZCRs were able to properly maintain bulk and spatial control within their normal operating ranges. The average zone level fluctuated between 57% and 67% with the maximum and minimum levels achieved by a zone being 85% and 32% respectively. The bundle and channel powers were maintained well below the limits with the maximums being 670 kW and 6660 kW. Single channel effects were measured on channel L12, the average zone levels were 4% less and the specific LZCR (zone 11) was 28% lower without fueling. The approximate reactivity insertion of the fuel into channel L12 is about 0.94 mk for the zone or 0.13 mk over the whole core (28% and 4% of the LZCR worth of 3.357 mk). A plot of the average and local zone 11 liquid zone level responses for the first day of fuelling simulation for both fuelling and not fuelling

A.C. Morreale et al. / Nuclear Engineering and Design 241 (2011) 3768–3776

3775

Fig. 8. Max bundle power distribution for days 200–390 of fueled operations. Fig. 7. Response of average and local liquid zone levels to fuelling.

channel L12 at 3 h is provided in Fig. 7. Although caution should be taken to avoid fuelling too much in one zone, the design is able to handle the reactivity inputs from fuelling. This fuelling simulation proved that the TRUMOX CANDU was able to handle day-to-day fuelling and maintain the prescribed limits. The next step in the fueling study was an extended operational simulation of 390 full power days (FPD) using the REFUEL code coupled with RFSP micro depletion simulations. REFUEL is an automated code that runs RFSP simulations for each full power day and fuels the core accordingly to maintain the reactivity level, simulating actual operations. The program chooses channels to be fueled using guidelines based on reactor characteristics such as LZCR levels channel ages, expected fueling ripple and bundle and channel powers. This automated process was run with the TRUMOX CANDU core for 390 full power days starting from one of the INSTANTAN snapshot cores.3 The long-term behaviour of the core including bundle and channel powers was recorded for this operation period. Maximum powers reached were 6.680 MW on FPD-343 in channel E15, and 873 kW on FPD-262 in channel L9 bundle 11. The average max channel power was 6.515 MW and the average max bundle power was 845 kW. The bundle powers display an upward trend in the first 150–200 days of about 10 kW (1.18% of average max BP), and over the full trial (0–390 days) 98% of the data is between 820 kW and 870 kW. At 390 FPD each channel has been fuelled at least twice and on average 3 times meaning that the channel has only experienced about 25% of its fuelling cycle. The initial 150–200 days of the simulation are dominated by the initial snapshot core from INSTANTAN producing the upward trend in bundle power described above. The REFUEL code takes some time to take full hold of the core and after about 200 days the data flattens out to a more level trend. Over the full 100% of the fuelling cycle the bundle powers will generally be stable about a median value. The max bundle power data for the period of stable fuelling (days 200–390) is shown in Fig. 8. The performance of the LZCRs during this extended fuelling trial was acceptable with the average levels staying between 46% and 68%. The average LZCR level during the 390 FPD trial was 55%. Some zones were stressed over the course of the trial but they tended to stay within the control envelope 78% of the time. The exception to this was zones 4 and 11, which had average levels of 4.7% and 4.8%. This resulted in lower powers in the inner region of the core, as it should have been fuelled more often. This problem could be

3 390FPD is sufficient for the full burnup cycle of NU fuel but due to the higher enrichment is not sufficiently long enough for the TRUMOX design to experience a full cycle. The preliminary nature of this design and computational limitations restricted efforts to extend the study further. Future detailed studies will evaluate the fuel with longer burnup histories.

remedied by modifying the fuelling rules to put more emphasis on the LZCR level and will be corrected as part of ongoing design and feasibility studies. A major finding in this phase was related to the fuelling strategy for the TRUMOX fuel. In particular, during portions of normal fuelling the reactivity inserted by fuelling caused some zone levels to depart from their normal operating envelope. It is unlikely that changes to the fuelling strategy (i.e. irradiation times, regions, bundle shifts, etc.) itself could overcome these effects. The TRUMOX fuel has a high fissile content in order to reach its burnup target of 45 MWd/kgHE and hence has a large reactivity insertion during fuelling. To compensate for the fuelling ripple caused by the large reactivity injection significant changes to the reactor control system may need to be implemented. Specifically, the effectiveness of the liquid zones could be increased by increasing their size or relative worth (i.e. the addition of neutron poisons to the LZCR water); however, the intent of this study was to avoid design changes. Alternatively the reactor design can be maintained and the fuel used can be altered. A fuel with a lower burnup target and hence a lower fissile content and reactivity insertion during fuelling could be used. This less reactive fuel would be more compatible with the current reactor control system and allow the LZCRs to stay within their existing prescribed envelope. A fuel with lower actinide concentration will have a lower enrichment, burnup level and fuelling ripple, reducing the effects of the zone levels during fuelling. For example, reducing the burnup target to 30 MWd/kgHE lowers the actinide concentration to about 3.1% (from 4.75%) and the fissile content to 2.53% (down from 3.5%). The reactivity impact from fuelling of this lower burnup fuel would be about 23.5% lower and the impact on the liquid zone levels would be reduced by a similar amount allowing the zones to better stay within their existing prescribed envelope and enhancing the controllability of the reactor. A significant outcome of this study is that the burnup target of 45 MWd/kgHE is too high given the CANDU control system and should be reassessed to a more reasonable level. The REFUEL simulation demonstrated the feasibility of long term fuelling with the TRUMOX CANDU design and did produce viable results. The results were similar to the time average but the center and some of the lower regions were not fuelled enough increasing the separation between the two models. The fuelling simulation can be brought closer to the time average by refining the fuelling rules and extending the simulation time to about 1500 FPD to encompass the full fuelling cycle. While the target burnup of 45 MWd/kgHE would be advantageous in terms of economic performance, it is likely that this will need to be relaxed and the effective actinide enrichment reduced in order for this type of fuel to be accommodated in existing operating reactors.

3776

A.C. Morreale et al. / Nuclear Engineering and Design 241 (2011) 3768–3776

4. Conclusions The use of a Transuranic Mixed Oxide fuel in a CANDU reactor was proven feasible and the control of such a core has been thoroughly evaluated. The fuel, known as TRUMOX, was a mixed oxide containing 95.25% natural Uranium blended with 4.75% actinides of Plutonium Americium, Curium and Neptunium. The actinides were reprocessed from spent Light Water Reactor fuel that had cooled for 30 years. The TRUMOX fuel was combined with the proven CANDU pressurized heavy water reactor platform. The TRUMOX CANDU design was able to attain 96.5% of the desired burnup target, 45 MWd/kgHE, while maintaining bundle and channel power limits of 935 kW and 7.3 MW at all times. However, during the fuelling studies some abnormal liquid zone levels were observed. This imposes potential limits on the enrichment/burnup targets of the TRUMOX fuel design. The model created proved to be controllable and safe under a wide variety of standard operational circumstances and required very minor changes to the CANDU platform. This reactor design will reduce fuel disposal costs and the increased costs of TRUMOX fuel can be offset by the efficiency gains of higher burnup. The use of reprocessed actinides increases the recycling and reuse of spent fuel producing a more efficient and sustainable fuel cycle. Acknowledgements The authors would like to thank the personnel at AECL Chalk River and Sheridan Park for their assistance with this research,

especially Bronwyn Hyland and Gary Dyck for the prototype fuel design, Peter Schwanke for the original CANDU RFSP model, and Cuong Ngo-Trong and Alan Gray for their help in reactor analysis methods. In addition, thanks are extended to Dr. Bill Garland and Dr. Mohammed Younis for arranging the research opportunity and to Dave Jenkins and Gary Dyck for supervising the project. Funding for this research was provided by AECL, UNENE and NSERC. References AECL, 1995. Fundamentals of Power Reactors – Module One: Science and Engineering Fundamentals [on-line]. Sec. 4., http://canteach.candu.org/library/ 20031205.pdf. Baetsle, L.H., De Raedt, C., 1997. Limitations of actinide recycle and fuel cycle consequences: a global analysis, Part 1: Global fuel cycle analysis. J. Nucl. Eng. Des. 168, 191–201. Collins, E.D., et al., 2004. Can thermal reactor recycle eliminate the need for multiple repositories. In: Proceedings of the 8th Information Exchange Meeting. Actinide and Fission Product Partitioning and Transportation, OECD Nuclear Energy Agency, Paris, France. Rouben, B., 2002. Introduction to Reactor Physics [on-line], pp. 79, 83, 131 http://canteach.candu.org/library/20040502.pdf. Taiwo, T.A., et al., 2007. Comparative study of plutonium burning in heavy and light water reactors. In: Proceedings of ICAPP 2007: International Congress on Advances in Nuclear Power Plants, Nice, France, May 13–18, 2007. Weigland, R.A., et al., 2007. Impact on geologic repository usage from limited actinide recycle in pressurized light water reactors. J. Nucl. Sci. Tech. 44 (3), 415–422. Wilson, P.D., 1996. The Nuclear Fuel Cycle: From Ore to Waste. Oxford University Press, Oxford, UK. Xing Guan, Z., et al., 1999. Physics codes and methods for CANDU reactor. J. Nucl. Pow. Eng. 1999-06.