In this study, analysis of the complete operational history of the “Joˇzef Stefan” Institute (JSI) TRIGA reactor was performed. Reactor power changes, core configurations and weekly excess reactivity measurements were analysed to obtain the needed data for fuel burnup calculations.
Trang 1Available online 23 October 2020
0149-1970/© 2020 The Authors Published by Elsevier Ltd This is an open access article under the CC BY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Anˇze Pungerˇciˇc , Duˇsan ˇCaliˇc , Luka Snoj
aReactor Physics Department, Joˇzef Stefan Institute, Jamova cesta 39, 1000, Ljubljana, Slovenia
bFaculty of Mathematics and Physics, University of Ljubljana, Jadranska ulica 19, 1000, Ljubljana, Slovenia
A R T I C L E I N F O
Keywords:
TRIGA research reactor and Fuel Burnup
Operational history analysis
in good agreement compared with the excess reactivity measurements Code-to-code comparison is presented Clear agreement is observed when comparing changes in core excess reactivity, and discrepancies are observed in the comparison of individual fuel element burnup and its isotopic composition The Serpent-2 results are in better agreement with the measurements compared to the TRIGLAV code; nevertheless, a conclusion can be made that the TRIGLAV code is viable for TRIGA fuel management and burnup analysis A three-dimensional (3D) burnup study was conducted, where individual fuel elements were further divided into multiple angular and axial depletion zones Notable burnup effects were observed, and an explanation using surrounding water distance is presented
1 Introduction
Determination of fuel element burnup in research reactors is an
important activity, as it is related to fuel utilisation and management,
characterisation of radiation fields in the reactor, reactor safety
pa-rameters and safety analyses, as well as spent fuel management The
“Joˇzef Stefan” Institute (JSI) has been operating a TRIGA Mark II reactor
(Douglas et al., 2003) since 1966 Over this time, multiple burnup
cal-culations and measurements have been performed (Ravnik et al., 1999;
ˇ
Zagar and Ravnik, 2000; Perˇsiˇc et al., 2000; Jeraj et al., 2002)
Mea-surements were performed using the fuel element reactivity worth
method (Ravnik et al., 1987) Burnup calculations for the JSI TRIGA
reactor were performed using only part of the operational history with
simplified operational data; this was done because the operational
his-tory analysis was not available Burnup was calculated with
determin-istic codes, such as the in-house developed TRIGLAV code (Jeraj et al.,
2002; Perˇsiˇc et al., 2017) TRIGLAV is a two-dimensional (2D) diffusion
code used for calculation of fuel element burnup, excess reactivity and
power peaking factors (Snoj and Ravnik, 2008) As the TRIGA core is
highly heterogeneous and asymmetric, three-dimensional (3D) Monte
Carlo codes are superior to diffusion codes for burnup calculations
Hence, we decided to repeat the analysis (reactivity worth of important
isotopes) using modern tools, which enable 3D Monte Carlo burnup calculations using detailed operational data Our goal was to improve the TRIGA burnup analysis using modern burnup codes and validate the results with multiple excess reactivity measurements The JSI TRIGA reactor had well-recorded individual reactor operation from 1966 to the present We decided to obtain the operational data and simulate the complete history using reactor physics and burn-up code Serpent-2 (Lepp¨anen et al., 2015; Maria, 2016) Only a few TRIGA reactors have this much information available regarding its operation (Chiesa et al.,
2016; Idris Lyric et al., 2013; Khan et al., 2010; El Bakkari et al., 2013)
In addition, two experiments (i.e criticality and fission rate profile) performed at the JSI TRIGA reactor have been published in the ICSBEP and IRPhEP handbooks (Jeraj and Ravnik, 2010) Several other exper-iments that can be used for validation of computer codes and models have been performed at the JSI TRIGA reactor (Raduloviˇc et al., 2014; Goriˇcanec et al., 2015; ˇStancar et al., 2018; Ambrozic et al., 2020) Our primary goal was to record and digitalise the complete opera-tional history, different core configurations and individual reactor op-erations for the purpose of experimental validation of computer codes and models As for the need for burnup experiments, we also decided to analyse weekly excess reactivity measurements The analysed opera-tional history is presented in the first part of the paper
* Corresponding author Reactor Physics Department, Joˇzef Stefan Institute, Jamova cesta 39, 1000, Ljubljana, Slovenia
E-mail addresses: anze.pungercic@ijs.si (A Pungerˇciˇc), luka.snoj@ijs.si (L Snoj)
https://doi.org/10.1016/j.pnucene.2020.103536
Received 22 June 2020; Received in revised form 29 September 2020; Accepted 9 October 2020
Trang 2Due to the significant amount of fuel shuffling throughout the history
and diverse reactor operation, we developed an automated script called
STRIGA (Pungerˇciˇc et al., 2019) This script creates a Serpent 2 input of
the TRIGA research reactor (Calic et al., 2016) based on the loading
scheme, fuel element composition and burnup parameters, and it
pre-pares necessary inputs for Monte Carlo burnup calculations This
methodology, together with that employed in the TRIGLAV code, is
presented in the second part of this paper The last part focuses on the
burnup calculations, comparison between both codes, and most
impor-tantly, the validation performed using the excess reactivity
measurements
2 JSI TRIGA operational history analysis
The first part of our detailed burnup analysis is the JSI TRIGA Mark II
operational history analysis, which consists of several important parts
that are presented in detail in this section These are as follows:
•Reactor description: description of the JSI TRIGA Mark II reactor
•Reactor power changes: analysis of individual reactor operations to
accurately calculate released energy
•Fuel shuffling: analysis of all of 240 core configurations used in the
complete operational history
•Measurements of excess reactivity: analysis of weekly
measure-ments used for validating burnup calculations
•Measurements uncertainty: evaluation of uncertainties in the
reactor operation parameters
2.1 Reactor description
Analysis of reactor operation for the purpose of detailed burnup
determination was performed on a 250 kW TRIGA Mark II research
reactor at the “Joˇzef Stefan” Institute Only a brief description of the
reactor is given here, comprising information that is important to
un-derstand the presented work; a more detailed description can be found
in descriptions of JSI TRIGA benchmark experiments (Raduloviˇc et al.,
2014; ˇStancar et al., 2018; Mele et al., 1994; Jeraj et al., 1997; Jeraj and
Ravnik, 2010; Ravnik and Jeraj, 2003)
The JSI TRIGA reached first criticality on 31st May 1966 Since then,
300 different fuel elements have been in use Information regarding all
different types of fuel element is presented in Table 1 In general, two
types of fuel elements were used as follows: stainless steel (SS) with a
zirconium rod in the middle and aluminium (AL) without the middle
rod These setups are presented in Fig 1 The SS fuel element features a
molybdenum disk below the fuel region, while the AL fuel element has a
samarium disk above and below In addition, the fuel elements are divided into three groups The first two groups have 8.5 wt % or 12 wt
% of 19.9 % low enriched uranium (LEU) in the U–Zr–H mixture, while the third one is 8.5 wt % of 70 % high enriched uranium (HEU) in the U–Zr–H–Er The latter is a so-called FLIP fuel element, and it contains erbium as a burnable absorber In this study the fresh fuel element composition was determined by considering recorded masses of 235U and 238U However, in 1999, all fuel elements except SS 12 wt % were shipped back to the United States, meaning that burnup calculation is the only way to determine their burnup and isotopic composition During the reconstruction of the JSI TRIGA reactor in 1991, the control rods were replaced, and also another different one (transient) was introduced Since then four control rods have been in use: P =transient, S = safety, R = regulating and C = shim The latter three feature a fueled follower SS 12% type fuel, which is thinner and has
r follower=1.66687 cm instead of r SS12 = 1.82245 cm, while for the
transient control rod, only air is left in its place when extracted The older control rods did not feature a fueled follower and were used in different core position Aluminium tube was left in their place when extracted In addition, the absorber for regulating control rod was
thinner, where r reg=1.1 cm instead of r shim,safe =1.6 cm A schematic
view of the control rods before and after the reconstruction in 1991 is presented in Fig 2
2.2 Reactor operation from 1966 to 2019
One of the most important parameters in burnup determination is the energy released during reactor operation Therefore, each operation written in the reactor logbooks was analysed In total, 50 logbooks or approximately 20 000 pages were analysed (depicted in Fig 3) Our goal was to digitalise all the needed parameters for future burnup calcula-tions The energy released can be calculated from the reactor power level, date and time of both reactor start-up and shut-down or power change A part of this information in a computer readable format is presented in Table 2
Using reactor operation data, the annual released energy in the reactor core is obtained (depicted in Fig 4) It can be seen that, after
1991, the energy released was reduced since the TRIGA reactor was no longer used for isotope production In recent years, the annual average reactor power decreased, mostly due to the higher number of reactor operations at lower power for detector testing, benchmark experiments and similar activities (Goriˇcanec et al., 2015; Raduloviˇc et al., 2014; Henry et al., 2015; Filliatre et al., 2015; ˇStancar and Snoj, 2017; ˇStancar
Table 1
Material composition of four types of fuel elements that were in use in the JSI
TRIGA Mark II research reactor (ˇZagar and Ravnik, 2000) For each fuel element
type, its years of utilisation are depicted
Fuel element name AL 8.5 % SS 8.5 % SS 12 % FLIP
aComposition of individual fresh fuel elements can vary but no more than 1 %
from the depicted values
Fig 1 Schematic view of the “Joˇzef Stefan” Institute (JSI) TRIGA fuel elements
with its dimensions Two different types of fuel elements based on cladding are presented: Stainless steel SS-304 (top) and aluminium AL (bottom)
Fig 2 Schematic representation of control rods used before (bottom) and after
(top) the reconstruction of the reactor in 1991 Two different types of old
control rods were in use, differing in the absorber radius Thicker (rabs =
1.6 cm) rod types were used for Shim (position C-03) and Safety (C-11) control rods, while a thinner (rabs= 1.1 cm) rod type was used for the Regulating
control rod (E− 13)
Trang 3et al., 2018) In addition the pulse mode operation started after the
reconstruction was analysed (Pungerˇciˇc and Snoj, 2018); however, the
energy released during the individual pulse is negligible due to the short
pulse duration ( ̃ 10 ms (Vavtar et al., 2020)); therefore it was not considered in the burnup calculations
2.3 Core configurations
Another important part of burnup calculation is the fuel shuffling history Throughout the history, the fuel element position in the core has been changed several times To acquire the positions, all core configu-rations (in total 240) were analysed and digitalised so their loading patterns could be used in the burnup calculations or for other activities
An example of two recent core configuration changes is presented in Fig 5
During the analysis of the reactor logbooks and the fuel shuffling, it was found that some fuel elements of the SS 8.5 wt % type were already previously used in another TRIGA reactor in Frankfurt am Main, Ger-many The problem was that the burnup of those fuel elements is un-known Therefore, additional analysis, as presented in 4.1.1, was performed to understand and evaluate the effect of not knowing the burnup of these fuel elements Another important discovery was that, after the reconstruction in 1991 new fresh fuel elements of the SS 12 wt
% type, which are still in use today, were mixed together with the old,
on 150 W, reactor power change to 250 kW and measurements of excess reactivity are depicted
Fig 4 Analysis of the JSI TRIGA Mark II reactor operation by each year from
its start in 1966 to the end of 2018 Annual released energy and average reactor
power are depicted for each year Average reactor power was calculated as
P avg=
∑
P i t i
t op , where i presents individual operation and top represents total time
reactor was operational in 1 year
Fig 5 Examples of three different core configurations of the JSI TRIGA reactor Each fuel element is labelled with a four-digit identification number All three core
configurations were established in 2018
Trang 4coupling the operational history before and after the reconstruction
This makes the burnup analysis more complex, as the complete history
from 1966 must be considered All the fuel shufflings are presented in
JSI TRIGA fuel shuffling animation (Reference to the online animation)
2.4 Measurements of excess reactivity
The last part of the operational history analysis was to record all the
excess reactivity measurements Excess reactivity has been determined
regularly every Monday since the start of the operation in 1966 Usually,
the JSI TRIGA reactor does not operate during the weekend, which
means that measurements are without xenon contribution Changes in
excess reactivity are directly connected to fuel burnup or fuel shuffling,
as presented in Fig 6 As these changes can be used to validate burnup
calculations, we have decided to analyse all 2000 measurements
per-formed in the complete operational history
2.5 Major sources of measurement uncertainty
A major source of uncertainty in the fuel burnup determination
re-lates to uncertainty in reactor power measurements In small research
reactors, the power is normally calibrated with respect to a single
neutron detector Its response is proportional to the flux at its position
Local flux is proportional to the total flux (power) of the reactor only if
its radial and axial distributions do not change However, this is not the
case in the research reactors where operational reactivity changes
(burnup, power defect, xenon effect) are compensated for by moving the
control rods Redistribution effects on neutron flux distribution due to
control rod insertion/withdrawal detected by a single detector may be in
the order of 20 % yielding the same error in reactor power readings
(Goriˇcanec et al., 2015; Zerovnik et al., 2014ˇ , 2015) Using two or more
detectors strategically located at different locations around the core can
reduce the error Recently, this uncertainty was reduced to 2 %–5 % with an improved thermal power calibration method (ˇStancar and Snoj,
2017) Other negligible uncertainty related to released energy is also in the reactor startup or power change as only time on power is written; therefore, the energy released during the transient is not considered Uncertainties in excess reactivity measurements are more difficult to evaluate since they depend on the changes in the reactor core For the comparison of absolute reactivity measurements, the 1σ uncertainty can
be up to 500 pcm due to the control rod-worth measurements (Trkov
et al., 1987; Merljak et al., 2018), reactor physical parameters (Filliatre
et al., 2015; Snoj et al., 2010; Henry et al., 2015) and power bution due to control rod insertion (Goriˇcanec et al., 2015) In the analysis of relative changes in excess reactivity, the assumed uncertainty
redistri-is much smaller as the same control rod-worth measurements are being used, and the changes in reactor physical parameters and flux redistri-bution are negligible if the core configuration is the same
3 Methods of burnup calculations
Until now, all burnup calculations for the JSI TRIGA Mark II, as presented in (Jeraj et al., 2002), have been performed either by using the TRIGLAV (Perˇsiˇc et al., 2017) fuel management deterministic code or other unit-cell based burnup calculations Usually the isotopic compo-sition was obtained with a standalone burnup code (e.g WIMS-D (Kulikowska, 1996)) and then used in Monte Carlo code, MCNP For the experiments with burned core configurations, higher discrepancies between the reactivity measurements and those calculated were observed (Zagar and Ravnik, 2000ˇ ) For this purpose, we have decided (in addition of using the TRIGLAV code) to simulate the complete operational history using the Monte Carlo neutron transport and burnup code Serpent-2 (Lepp¨anen et al., 2015), which is known for its burnup capability (Maria, 2016) Nuclear data libraries for both codes are based
on the ENDF/B-VII.1 (Chadwick et al., 2011) nuclear data Comparison
of continuous (Serpent) and 69-group cross-section (WIMS-D) energy dependence for total neutron incident on 235U is presented in Fig 7
3.1 The TRIGLAV code
The TRIGLAV fuel management and burnup code was developed at the Reactor Physics Department of the “Jozef Stefan” Institute A general description of the code is given in (Perˇsiˇc et al., 2017), and a detailed description of the burnup calculation is provided in (Ravnik et al.,
1999) The code is based on a four-group diffusion equation for r − ϕ
geometry, solved by the finite difference method The TRIGLAV gram package consists of a four group 2D diffusion module and the WIMSD-5B code (Kulikowska, 1996), which is linked automatically to the diffusion module to calculate unit-cell-averaged effective group constants All 91 positions in the reactor core are treated separately in the unit-cell approximation The TRIGLAV geometry model, presented
pro-in Fig 8, represents the full TRIGA cylindrical core and graphite
Fig 7 Energy dependence for total 235U microscopic cross-section Continuous
(Serpent-2) and 69-group (WIMS) cross-sections are presented The latter is
acquired from the WIMS Library Update Project (Coordinated Research), and it
is based on the ENDF/B-VIII.1 nuclear data library (Chadwick et al., 2011)
Fig 6 Measurements of excess reactivity performed at the JSI TRIGA reactor in the complete operational history (left) and selected period of 9 months (right),
where changes due to core shuffling and burnup are visible Four core configuration changes were employed in this period Energy released during operation on core configurations 30 and 32 is depicted
Trang 5reflector On the reflector outer boundary, the zero flux boundary
con-dition is imposed The unit-cell-averaged cross-sections were calculated
using 69-group WIMS library based on the ENDF/B-VII.1 nuclear data
library
Fuel element burnup (BUel) within the TRIGLAV code is calculated
using the WIMSD-5B code from which the fuel isotopic configuration is
obtained, as seen in Fig 8 Unit-cell homogenised cross-sections at the
defined burnup are extracted and collapsed into neutron flux weighted
four-group constants that are used in 2D diffusion approximation With
diffusion solution, a core power distribution is obtained and the fuel
element burnup increment can be calculated by knowing energy
released data for the defined cycle Using the described procedure, complete operational history can be simulated where different core configurations and reactor operations can be considered In the current TRIGLAV methodology, the isotopic composition of individual fuel el-ements is not transferred from cycle to cycle as only the burnup incre-ment for individual fuel elements on each cycle is calculated Such procedure means that the mentioned BUel carries the information regarding the complete operational history (power and loading pattern changes)
Fig 8 TRIGA Mark II reactor geometry in the TRIGLAV model with homogeneous unit-cells (left) TRIGLAV code methodology schematic flow-chart (right)
Fig 9 Schematic representation of the STRIGA methodology in which TRIGA reactor parameters are used to create Serpent-2 input for steady-state or burnup
calculations Each of the four important inputs is described in the schematic
Fig 10 Graphical representation of the TRIGA Mark II reactor diverse power operation treatment in the simulation of the complete operational history
Trang 63.2 STRIGA tool
The STRIGA tool is a simple data manipulation script, written in
standard FORTRAN-77 programming language that reads the TRIGLAV
inputs and creates inputs for 3D Monte Carlo calculations Core
configuration inputs in TRIGLAV were already prepared for the
com-plete operational history Detailed description of the STRIGA tool is
presented in (Pungerˇciˇc et al., 2019; Caliˇc et al., 2017ˇ ); only a brief
description is presented in this paper The STRIGA tool is based on the
validated steady-state Serpent-2 model (Calic et al., 2016), that is based
on the criticality benchmark (Mele et al., 1994) and MCNP models used
in (Raduloviˇc et al., 2014; ˇStancar et al., 2018; Goriˇcanec et al., 2015)
Steady-state SERPENT-2 calculations are completely consistent with the
MCNP ones, indicating that the geometry and material composition
employed in the model are well defined for reactor cores with fresh fuel
(Calic et al., 2016)
STRIGA requires two kinds of input data The first represents reactor
component dimension and its material composition, while the second
represents the reactor operation data, as depicted in Fig 9 To simulate
the complete history from 1966 to 2019, additional types of fuel
ele-ments and control rod positions have been added to the existing STRIGA
script (Caliˇc et al., 2017ˇ ) Another advantage of the STRIGA tool is that it
reads the TRIGLAV core configuration input file, which is created using
the graphical user interface (GUI) Triglav-W (ˇZagar et al., 2006)
Through this interface, one can specify the reactor operation (power and
time) and select the core loading patter via an user-friendly fuel
shuf-fling interface All fuel and non-fuel elements can be moved from one
location to another by a simple click and point procedure In addition,
cool-down of the reactor core was added into the STRIGA tool, enabling
detailed calculations of diverse reactor operations
For a given fuel cycle calculations using burned fuel the most
important information is the fuel isotopic composition that is taken from
the previous cycle Within the STRIGA tool, the user can select which
isotopes are tracked within the burnup calculation In the analysis, 269
isotopes were tracked After each burnup calculation the fuel isotopic
library is created or updated Such principle enables the extraction of
individual fuel element isotopic composition at different times in the
reactor operational history
4 Burned fuel material composition calculations
Fuel element burnup analysis for the TRIGA Mark II reactor is
per-formed on three different cases, which are as follows
•Complete operational history: Analysed data regarding the energy
released and fuel shuffling was used to create a burnup simulation of
the actual reactor operation performed from 1966 to 2019
•Burnup on benchmark core No 132: Burnup simulation on
selected core configurations was performed to study the differences
between the Serpent-2 and the TRIGLAV code Criticality benchmark core 132 (Jeraj et al., 1997; Mele et al., 1994) was selected
• Burnup on fuel unit-cell: Burnup simulation on fuel element unit-
cell (fuel pin surrounded with water) to study the physics of fuel composition changes through burnup
4.1 Simulation of complete operational history
The complete operational history consists of more than 25 000 reactor power changes and more than 240 core configurations To simulate each individual operation, high computer power and memory would be needed, especially for calculations with Monte Carlo code Serpent-2 Therefore, additional simplification was used to divide the complete operational history into individual long operations on same loading pattern or core configuration, as presented in Fig 10 It has been
approximated that the reactor operated on maximum power Pmax=
250 kW for a period in which the energy released
∫t
0Preactor(t)dt was the
same At the end of the operation, fuel cooldown was approximated as the sum of the total time the reactor was not operational This principle was used to create 240 different inputs for both Serpent-2 and the TRIGLAV code In this way, individual fuel element burnup was tracked from its first insertion in the reactor core until today
The approximation that the reactor operated on maximum power
Pmax=250 kW was tested by repeating burnup simulation on a thetical core configuration using the Serpent-2 code Calculations were repeated for 100 kW and 1 kW reactor powers To keep fuel burnup the same at different reactor powers, the irradiation time was increased accordingly The differences at the point of maximum TRIGA fuel burnup were less than 0.1 % and 1 % for differences in the calculated
hypo-nU− 235 and nPu− 239, respectively We also repeated the simulation for
Preactor=100 W and found similar discrepancy for nU− 235 and higher 10
% discrepancy for nPu− 239 Reason behind higher relative differences in
nPu− 239 is that the amount of 239P in TRIGA fuel elements is low, due to
the smaller content of 238U compared with traditional light-water
re-actors Relative differences in nU− 235 and nPu− 239 are presented in Fig 11 Despite higher relative differences observed, this effect is negligible when comparing final absolute values in material composi-tion Similar conclusion can be made for fuel and moderator tempera-
ture We have investigated the effects of Tfuel and Twater by repeating the
calculations with Tfuel=600 K and Twater =350 K It has to be noted
that Twater represents the temperature of the water surrounding the fuel and it is only one of the contributions to the neutron moderation in a TRIGA reactor The density of the water surrounding the fuel was
Fig 11 Differences in calculated nU− 235 (left) and nPu− 239 as a function of fuel element burnup for different reactor powers defined in the burnup simulation Pmax=
250 kW was used as a reference
Trang 7nU− 235 and nPu− 239, respectively Resuls are presented in Fig 12 The
noticeable effect on plutonium production is constant at 10 % This is
due to the Doppler broadening of (n, γ) reaction resonances on 238U
There were no noticeable effects of surrounding water temperature on
nPu− 239 and for nPu− 239 the differences were below 0.1 % A conclusion
can be made that for absolute determination of 239Pu, fuel temperature
has to be taken into account, however for our analysis, mainly consisting
of comparing relative change due to burnup, negligible discrepancy is
introduced However further more detailed analysis with detailed
thermodynamical model should be performed to fully analyse the
tem-perature effect on burnup
Analysing the effect of different reactor powers and fuel
tempera-tures on depletion of 235U and production of 239P showed that there is
negligible effect on core excess reactivity due to long-lived isotopes if
approximation presented in Fig 10 is taken into account In addition we
observed that 239P production depends slightly more on different reactor
power than fuel temperature, especially in smaller burnup below
kg(HM) However the amount of 239P produced in TRIGA fuel is low and
such effects could be neglected in some cases However effect on core
excess reactivity is present due to short-lived isotopes such as xenon and
samarium Effect of such isotopes is analysed in Sec 6.1 From this a
conclusion can be made that if xenon and samarium are of interest,
detailed operational history for the last couple of days or weeks should
be taken into account
Burnup of all 300 fuel elements accumulated in different periods of the reactor operation history from 1966 to 2019 was calculated with both the TRIGLAV and the Serpent-2 code Complete operational history simulation with TRIGLAV takes approximately 1 h on one standard PC core, while Serpent simulation takes approximately three weeks on 56 cores (Intel Xeon Processor 25M Cache, 2.80 GHz) Accuracy of the calculated burnup depends mainly on the experimental power calibra-tion accuracy (ˇStancar and Snoj, 2017) and the precision of the opera-tional records As discussed in section 2.5, the 1σ uncertainty in reactor power level is 10 %–15 % However, we assumed only 5 % uncertainty
in final burnup as the uncertainty for longer operations averages out The final fuel burnup for randomly selected 16 fuel elements, 4 of each type, (each fuel element is represented with a four-digit fuel identifi-cation number) is presented in Fig 13 It should be noted that these fuel elements were not part of the same core configuration, but they were used in different parts of the operational history
First, we can observe that FLIP fuel elements have higher burnup (up
to 70 MWd
kg(HM)),1 while the burnup for LEU fuel elements is around
Table 3
Material information defined in the Serpent-2 model of the JSI TRIGA reactor for
the analysis of fuel and water temperature effects on the fuel element burnup
Analysed cases Tfuel [K] ρfuel[ g
Fuel temperature test 600 6.04498E+00 300 9.97245E-01
Water temperature test 300 6.04498E+00 350 9.73742E-01
Fig 13 Final fuel burnup for 16 out of 300 fuel elements used in the history of
the JSI TRIGA reactor operation and calculated by simulating the complete
operational history For each fuel type, 4 elements were chosen, which are
represented with a four-digit identification number, with exception of fuel
followers for regulation (REGU) and safety (SAFE) control rods
Fig 14 Relative difference in calculated fuel burnup between Serpent and
TRIGLAV as a function of burnup for all 300 fuel elements used in JSI TRIGA operation Relative difference is defined as ( BU Serpent − BU TRIGLAV )
BU Serpent
Fig 12 Differences in calculated nU− 235 (left) and nPu− 239 as a function of fuel element burnup for higher fuel temperature Tfuel=600 K and core water temperature
Twater=350 K Tfuel=Twater=300 K was used as a reference and for all burnup calculations presented in this paper Both calculations were performed on maximum
steady state power Pmax=250 kW
1 Standard unit for burnup is energy released in MWd per mass of heavy materials (HM) in kg Heavy materials are nuclei with more than 92 protons
Trang 830 MWd
kg(HM), due to the higher local power of the FLIP fuel The difference
in calc burnup between TRIGLAV and Serpent is due to the higher
thermal neutron flux (Snoj and Ravnik, 2008) in the FLIP fuel elements
Furthermore, the discrepancies between the codes are analysed
Comparing the absolute differences, we can observe higher
discrep-ancies for FLIP-type fuel elements and the control rod fueled followers,
depicted as REGU for regulating and SAFE for safety control rod For
FLIP-type fuel elements, the discrepancies can be explained similarly to
the difference in calculated burnup Higher localised neutron flux results
in higher neutron flux gradients, which are not handled well by the 2D
diffusion approximation employed in the TRIGLAV code Nevertheless,
the differences in fuel burnup are within 10 % For fueled followers, the
relative difference is up to 15 % and this is because in the TRIGLAV code
we assume that the fueled follower is the same as a normal SS 12 % fuel
element, while in reality, it is smaller; this is considered in the Serpent-2
model, as previously presented in Fig 2
For all 300 fuel elements, relative difference in calculated final
burnup with both codes was evaluated The highest discrepancies were
observed for control rod fuel followers and for the fuel element type SS
8.5 % with unknown initial burnup (obtained from TRIGA in Frankfurt
am Main, Germany) The difference for the SS 12 % type of fuel, which is
most important from the standpoint of the current reactor operation, is
within the ±7.5 % range We can conclude that the difference between
both codes for most of the fuel elements is satisfactory, within 5 %, accounting for all the differences between the two codes Code-to-code comparison for all 300 fuel elements is presented in Fig 14
To compare the differences between fuel elements of the same type, core configurations and their position in the reactor core becomes important For this purpose, the following three core configurations were chosen to further analyse the calculated burnup and the differences between TRIGLAV and Serpent-2:
• Core No 80: due to mixture of three different types of fuel elements
(SS, AL 8.5% and FLIP)
• Core No 130: as it is the last core that was in operation before the
reconstruction in 1991 (maximum burnup in some fuel elements)
• Core No 240: as being in operation when the study was performed
kg(HM)for all types of fuel
Fig 15 Individual fuel element burnup calculated by considering the complete operational history Out of 240 core configurations, three were chosen (from left to
right: 80, 130, 240), schematically presented at the top For each configuration fuel burnup calculated with Serpent at the end of the cycle is presented together with relative differences between both codes used, defined as ( BU Serpent − BU TRIGLAV )
BU Serpent
Trang 9elements For core no 130, three clear sets of fuel elements are visible:
The first is FLIP, with the highest burnup up to 70 MWd
kg(HM); the second is SS 8.5 % with burnups between 10 MWd
kg(HM)and 20 MWd
kg(HM); and the last is SS 8.5
%, which were brought from the Frankfurt TRIGA reactor and had
recorded burnups below 10 MWd
kg(HM) The effect of burnup is clearly visible for core no 240, where burnups in the middle part of the core are higher,
at up to 20 MWd
kg(HM), compared with those on the periphery (up to 15 MWd
kg(HM))
Fuel element burnup results are presented in top part of Fig 15, while in
the bottom part, relative differences between Serpent-2 and TRIGLAV
for the same core configurations are presented A slight shift of
calcu-lated burnup can be observed when analysing differences between both
codes for the complete core, as the burnup calculated with TRIGLAV is
higher compared with Serpent-2 on the right side of the core, which can
be seen in Fig 15 This difference may occur due to the control rod-
induced neutron flux redistribution (Goriˇcanec et al., 2015) or the
treatment of burnup in TRIGLAV To understand the mentioned shift,
further analysis on just one core configuration was performed; this is
presented in section 4.2
In addition to burnup, isotopic composition of individual fuel
ele-ments was analysed Based on an analysis of isotopic effect on reactivity
(Jeraj et al., 2002), only selected isotopes were chosen In Fig 16, the calculated number densities for two important isotopes in burnup analysis 239P and 137Cs, are presented The former is the product of
neutron absorption in 238U, and as expected, the calculated amount is lowest in the HEU FLIP fuel elements The latter is the product of nuclear fission; therefore, similar behaviour compared with fuel burnup can be observed For fuel element with ID no 288, the lowest number densities for both isotopes can be observed The reason is that, in this experiment, this fuel element contained only a quarter of the fuel compared with the others Higher discrepancies are observed between the two codes with increased burnup For 235U, 137Cs and 239Pu analysis of the calculated number density differences between both codes was performed for core
No 240 The results are presented in Fig 17 For 235U and 137Cs, similar behaviour to fuel burnup can be observed, with the highest discrep-ancies observed for control rod fuel followers The radial neutron flux distribution, calculated with both codes is highly visible when analysing
239Pu as for the inner rings (B, C and D) higher number density up to 10
% is calculated with Serpent-2 and lower up to 15 % for outer rings (E and F) compared with the TRIGLAV code
The calculated differences for most fuel elements were within 5 %,
Fig 17 Comparison of calculated isotopic composition between Serpent and TRIGLAV for last core configuration established in 2019 Relative difference is defined
as NSerpent −NTRIGLAV
NSerpent Three isotopes are presented; from left to right, these are: 235U, 239Pu and 137Cs
Fig 16 Final 239Pu (top) and 137Cs (bottom) number densities for 16 fuel elements calculated by simulating complete operational history It should be noted that aluminium fuel element No 288 has only a quarter of fuel material compared with other AL 8.5 % elements
Trang 10while higher discrepancies were due to difference in control rod
treat-ment in both codes The difference in methodology for calculating
neutron flux distribution for both codes comes into effect when
comparing distribution of different isotopes, especially 239Pu These
differences are investigated further in the next chapters We can
conclude that the difference in the calculated burnup between both
codes is relatively low Hence, both methodologies could be used for
TRIGA fuel elements’ burnup calculations However, the user should be
aware that these differences increase for individual isotope density
comparison
4.1.1 Uncertainty propagation of initial fuel element burnup
In 1980, already irradiated fuel elements were obtained from the
Frankfurt TRIGA reactor As the initial burnup of these fuel elements was
not accurately known, additional uncertainty was introduced in the final
burnup calculations The effect of this uncertainty was studied, and the
results are presented in this section The STRIGA tool was used to study
the propagation throughout different core configurations
In 1991, soon after the reconstruction, these irradiated fuel elements
of type SS 8.5 % were used in mixed-core configurations from core No
138 to core No 146 (reference to the online animation) with the fresh SS
12% Their effect is studied by analysing the burnup of the fresh fuel that
is present in the mixed cores In total, fuel was mixed in multiple
different cores, which were studied in our case We have chosen 20 MWd
kg(HM)
as a reference value of the unknown fuel burnup This burnup was later
changed from − 100 % to +30 % and effects on accumulated burnup on
the fresh fuel elements were analysed The effect on final burnup after
core No 146 due to the mentioned burnup changes in a mixed core
configuration with fresh fuel is less than 6 % This effect is reduced for
further operation and it is practically negligible for fuel burnups at core
No 240 The effect on core No 146 is presented in Fig 18
4.2 Burnup on benchmark core no 132
The complete operational history includes a large number of
different parameters that could impact the calculated burnup, such as
fuel shuffling or mixing of different types of fuel elements To analyse
the burnup in higher detail and compare the two methodologies used we
decided to perform burnup calculations on the benchmark core
config-uration with fresh fuel no 132 (Jeraj et al., 1997; Mele et al., 1994) The
core schematic is presented in Fig 19 An average core burnup of 50
MWd
approximately represents the energy released in the complete
opera-tional history, which is an overestimation for only one type of fuel
element, while the core burnup of 20 MWd
kg(HM) represents the energy released after the reconstruction in 1991 and can be used as a realistic
example
The validation of the computational model with fresh fuel was
already performed (Calic et al., 2016) and taken as an initial condition in
this analysis Excess reactivity as a function of core burnup was studied
Both codes predict a similar change of ≈ − 6000 pcm at 20 MWd
kg(HM)and ≈
− 13000 pcm at 50 MWd
kg(HM) Both codes are in good agreement as the
discrepancy gradually increases to a difference of only 784 ± 20 pcm at
50 MWd
kg(HM) Higher discrepancies are observed for the first couple of steps
because of the xenon equilibrium treatment employed in the TRIGLAV
code Comparison of reactivity changes with both codes is presented in
Fig 20
The individual fuel element burnup and its isotopic composition was
studied The results are presented for three different core burnups,
which are as follows: 10 MWd
kg(HM) , 20 MWd
kg(HM) Serpent- calculated burnup parameters and discrepancies in comparison with
the TRIGLAV code are presented in Figs 21 and 22 As expected, the fuel
burnup is higher in the inner rings and becomes lower towards the
outermost ring This difference is almost negligible at lower burnups of
10 MWd
between fuel element in ring A and E is 20 MWd
kg(HM) Similar behaviour can
be observed when analysing the production of isotope 239Pu
Differences in fuel burnup and consequently individual isotopes between both codes were analysed Serpent results were taken as a reference value Highest discrepancies, similar to operational history analysis, are observed for control rod fueled followers and the central
Fig 19 Schematic representation of core configuration No 132, used as a
fresh fuel criticality benchmark and available in the ICSBEP handbook (Jeraj
et al., 1997; Mele et al., 1994)
Fig 20 Excess reactivity (top) and difference between Serpent and TRIGLAV
criticality calculation (bottom) as a function of TRIGA core No 132 burnup
Fig 18 Relative change in burnup (y-axis) of initially fresh fuel elements (blue
dots) due to uncertainty in initial burnup of already irradiated fuel elements, expressed as a relative change from the reference value (x-axis) (For inter-pretation of the references to colour in this figure legend, the reader is referred
to the Web version of this article.)