The MIRTE (Materials in Interacting and Reflecting configurations, all Thicknesses) program is an experimental program intended to study the influence on criticality hazards of specific structural materials arranged around and between fissile material elements. The experimental results contribute to the validation of the calculation of these materials using criticality softwarecalculation package. The program, which comprises a total of 115 experiments, was conducted in two phases: the MIRTE 1 experiments initiated in 2008 were completed in 2010; the MIRTE 2 experiments started up in January 2011 and were completed in December 2013.
IRSN, by virtue of its main vocation of improving safety, aims at ensuring the reliability of criticality safety margins determined by calculation for all types of configurations and, in particular, for those containing structural materials. In fact, for several years, calculations performed to justify the sub-criticality for basic nuclear facilities (INB in French) and transport packages have been considering configurations that are increasingly realistic by modelling in particular the materials present around the fissile material. These materials, given their absorption of neutrons, are in some cases decisive in ensuring sub-criticality to the extent that they decrease reactivity. It is thus vital to obtain qualification of these materials in order to correctly estimate the safety margins.
Criticality hazards are evaluated using criticality calculation packages that estimate the effective multiplication factor of neutrons (keff), in other words the ratio of the number of neutrons produced over the number of neutrons lost in the system. The purpose of the MIRTE program is to assess the discrepancies (calculation bias) between the keff calculated by the criticality calculation package and the experimental keff for the different configurations involving the structural materials of interest. Depending on the value of these discrepancies, a calculation bias may be considered in the safety margins.
The MIRTE program, which has been carried out in CEA’s Valduc Apparatus B experimental facility in France, was initiated in 2005 by IRSN and, in 2007, evolved into an international collaboration including the Areva Group, Andra and the US Department of Energy (DOE). The experiments, strictly speaking, began in 2008 and the program entered its second phase in January 2011 with MIRTE 2, involving the French industrial partners AREVA and ANDRA.
The program was carried out in two phases: MIRTE 1 followed by MIRTE 2. MIRTE 1 involved 43 experiments conducted between the end of 2008 and June 2010. The effect of nine materials - aluminum, copper, iron, nickel, lead, zircaloy, titanium, glass (SiO2) and concrete with different water contents (3, 6 and 9%) – was tested and, for each configuration, a reference experiment without material was carried out.
MIRTE 2, which enabled other materials to be tested, involved 73 experiments conducted between January 2011 and December 2013. MIRTE 2 was conducted in three phases: MIRTE 2.1, to improve the accuracy of some of the experiments from MIRTE 1 and to understand the discrepancies observed with the criticality calculation packages, MIRTE 2.2 to test new materials in the same configurations as MIRTE 1, and MIRTE 2.3 implementing new configurations with iron or copper sleeves to optimize feedback on the nuclear data by increasing the sensitivity of the keff to this data.
The experimental configuration is constituted of a stainless steel support structure (see Figure 1) containing arrays of fuel pins and the structural materials to be tested that interact with them. A measuring probe that monitors the rise of the water level in the experimental vessel is installed within this structure and makes it possible to follow the progress of the sub-critical approach and to determine when criticality is going to be reached. For each water measurement level, the neutron population is determined by at least three BF3 counters. The inverse of the neutron count rate as a function of the water height is extrapolated to zero, which determines the critical water height. The materials chosen are tested in the form of screens of different thicknesses and the experimental conditions are determined in order to end up with an overall experimental uncertainty below 0.2 %.
Figure 1. Experimental device used in MIRTE program © IRSN
The experimental configuration has been designed to enable easy modifications and replacements of the materials in place (see Figure 2).
Figure 2. Schematic views of the experimental configuration © IRSN
The fuel pins used in the MIRTE experiments are the UO2 pins available in CEA’s Valduc Apparatus B experimental facility. Their dimensional and chemical characteristics are thus well known; the 90-cm-high fuel column comprises a stack of 1.495-cm-high pellets, composed of uranium slightly enriched in 235U and clad with Zircaloy-4. Each fuel pellet has a radius of 0.395 cm and is surrounded by a cladding of external radius equal to 0.475 cm and thickness equal to 0.057 cm.
Description of the project and methodology
The experiments consisted in a sub-critical approach by progressively raising the water level (moderator and reflector) in the the Apparatus B experimental vessel. The experiment stopped when the effective multiplication factor obtained was equal to (1.00000 – ß/10), where ß represents the proportion of delayed neutrons. For this type of pins array, ß was of the order of 700 pcm (i.e. 0.7 %), which means that the criticality threshold was approached to some 70 pcm.
These experiments involved the use of one or more UO2 fuel pin arrays in which the uranium was enriched to 4.738 % in 235U. Three types of configurations were studied:
- "Reflecting" configurations: the experiments consisted in testing the reactivity effect of blocks of aluminium or glass surrounding the four faces of an array of fuel pins.
- "High thickness interacting" configurations: the experiments consisted in studying the effect on reactivity of a block of iron, nickel, zirconium, aluminium, copper or concrete (with water contents ranging from 3 to 9 %) of thickness between 5 cm and 30 cm, inserted between two arrays of fuel pins.
- "Low thickness interacting" configurations: the aim of these experiments was to characterize the effect on reactivity of the insertion of cruciform screens made of copper, nickel, iron or titanium of thickness below 20 mm, separating four arrays of fuel pins.
For these experiments, the anti-reactivity worth of the material tested varied from 4000 pcm to 10000 pcm.
In addition, in order to make it possible to evaluate any calculation bias due to the structural materials tested, reference experiments - identical configurations without material – were carried out. These experiments were optimised thanks to the use of the sensitivity calculation sequence, Tsunami-3D, of the Scale 5.1 calculation package.
Furthermore, 15 reproducibility experiments (a new sub-critical approach was carried out for the same experimental configuration) made it possible in particular to estimate, other than by calculation, the positioning uncertainties of the whole experimental set up (position of the pins, material screens, etc.).
In total, 43 sub-critical approaches were carried out (see table 1).
Table 1. MIRTE 1 experiments: configurations and screen dimensions © IRSN
Then, the experiments were compared with five criticality calculation packages associated with three nuclear data libraries. The two software packages using formalism implementing a multigroup approximation of nuclear data (APOLLO2-MORET 4 developed by CEA and IRSN and KENO-V.a, American code) have short calculation times because they use average values of the nuclear data contained in the libraries and they have a tendency, depending on the calculation route used, to overestimate (APOLLO2-MORET 4) or under-estimate (KENO-V.a) the criticality hazards. On the other hand, the three software packages using non-averaged values (MORET 5 developed by IRSN, TRIPOLI-4 developed by CEA and the American software MCNPX 2.6) give keff values closer to those of the experiments.
The influence of the libraries was also determined by comparing the values of the keff calculated by a software calculation package without approximation for different nuclear data libraries. For some materials, such as titanium, the nuclear data do not influence the effective multiplication factor value (keff), unlike lead, zirconium, silicon or aluminium, for which significant differences appear depending on the library used.
first phase of MIRTE 2, MIRTE 2.1, which started in January 2011, consisted in refining certain results of MIRTE 1 and understanding the trends identified when exploiting these results.
Certain reproducibility experiments (with copper and titanium) highlighted a possible effect of sampling of the batch of pins. It seems that the choice of the batch of pins has a quantifiable impact on the value of the critical height. Experiments testing different batches of pins selected randomly from the 1260 ones available lent weight to this observation. However, the effect remains negligible with respect to all of the experimental uncertainties, and it was not deemed necessary to identify each rod and its position within the rod array for future programs.
Lastly, MIRTE 1 seemed to show a potential effect of the position of the neutron counters. Depending on the position of the BF3 counters and depending on their sensitivity, a difference of few millimeters was observed on the critical height of the water. To better understand the effect and to quantify it to provide feedback for the existing experimental configurations, neutron counter placement experiments were conducted in MIRTE 2.1. The results of these experiments confirmed an effect of the position of counters for the most compact configurations, when these counters are close to the last peripheral rods in the array.
The second phase of MIRTE 2, MIRTE 2.2, conducted between 2012 and 2013, involved new materials - namely magnesium, molybdenum, chromium, rhodium, chloride and industrial resins BORA, VYAL B - using the same experimental device as in MIRTE 1. They are tested only in interaction-type configurations. The aim was to propose experiments that represented industrial configurations in terms of both configuration type and sensitivity profile to nuclear data. The sizes of the arrays and the thicknesses of materials were determined using the APOLLO2-MORET 4 and TRIPOLI-4 codes using the JEF2.2 nuclear data library. In addition, the Tsunami-3D sequence of the American SCALE package was used to optimize sensitivity to nuclear data.
Seven configurations were thus chosen. No new reference experiment was needed given that those from the MIRTE 1 phase could be reused. However, many reproducibility experiments were conducted to find out more about the experimental uncertainties
(see Table 2).
Table 2. MIRTE 2.2 configurations © IRSN
The third phase of MIRTE 2, MIRTE 2.3, took place at the end of 2013 with the aim of testing other configurations requiring adjustments to the experimental system. It was therefore decided to test the materials in sleeves placed around the fuel rods to increase the thermal spectrum sensitivity of the keff to the nuclear data of the material. In addition, configurations using a block of aluminum pierced with holes holding the clad rods with the material to be tested were produced. Reference configurations (without sleeves around rods and without aluminum blocks) and reproducibility experiments were also conducted.
The configuration thus comprises:
- a central zone comprising 400 rods surounding or not with iron or copper sleeves, of 19mm and 14mm thickness respectively; the latter may also be included in an aluminum block (see photo) ;
- a driver zone surrounding the test zone, comprising the same rods.
MIRTE 2.3: Rods in their copper sleeves (on right)
and aluminium block in the test zone before placement
of rods (on left)
The list of experiments is given in the table below.
MIRTE 2.3 configurations © IRSN
As for the MIRTE 1 program, the experiments were calculated with the five criticality calculation packages associated with the three nuclear data libraries mentioned previously. Analysis of the results revealed the discrepancies associated with certain approximations from the multi-group calculation route as well as nuclear data effects.