The safe operation of LWR NPPs requires that the integrity of the reactor pressure vessel (RPV)—which is endangered by neutron-induced embrittlement—be guaranteed during the whole lifetime of the plant. Reactor dosimetry is an important field since it provides the neutron fluence data that are used for the evaluation of the material irradiation damage, and therefore it is a crucial input for the safety assessment of any nuclear reactor. Plant life management needs reliable estimation of radiation field parameters (and their uncertainty) for reduction of the conservatism of material damage assessment and RPV lifetime estimations.
Radiation monitoring is an approach for non-destructive determination of neutron exposure of the RPV, its internals, claddings, and prediction of the radiation damage. Thus reactor dosimetry provides major input for the safety evaluation of the reactor.
Reactor dosimetry is needed to study a material’s stability, to allow prediction of damage, to quantify material damage, to assess/verify safety margins for plant life management (PLIM), to evaluate the remaining RPV lifetime, and to support PLIM strategies. The importance of these issues requires that improvement of the accuracy and reliability of the RPV fluence determination be a continuous task.
RPV neutron fluence is calculated. The main methods used in the countries operating VVERs are the TORT code’s discrete ordinate method [1] and MCNP code’s the Monte Carlo method [2]. The TORT code is used with the problem-oriented neutron cross sections library BUGLE96 [3] for PWR type reactors, and BGL [4] for VVER type reactors. The MCNP code is applied with the neutron cross section DLC200 library which represents the energy dependence of the cross sections continuously. The MCNP approach is considered a reference method because of the use of fewer approximations in neutron transport modeling.
The reliable calculation of neutron source (power and fuel burn-up) distribution in the core is a very important part of the fluence evaluation methodology. The precise fast neutron source distribution and its time evolution in the core periphery are especially important because this source gives the greatest contribution to the RPV fluence.
Careful benchmarking and development of new experimental and calculation methodologies for accurate determination of neutron fluence and improving the reactor dosimetry of VVERs were carried out by different laboratories. Full-scale mock-ups of VVER-1000 [5] and VVER-440 reactors [6] including core, internals, pressure vessel and biological shielding were developed to study and assess the reliability of the calculation approaches as well as to create benchmarks for verification of the calculation methods. These mock-ups were developed on the LR0 reactor of the Nuclear Research Institute in Řež, near Prague, in collaboration with RRC Kurchatov Institute (Moscow, Russia) and Škoda Nuclear Machinery (Plzen, Czech Republic). Experiments have been carried out for more than twenty years. The experimental results obtained are state-of-the-art measurements of neutron and gamma field parameters. The results refer to three mock-ups: mock-up of VVER-1000, mock-up of VVER-440 with standard core loading pattern, and mock-up of VVER-440 reactor with dummy cassettes instead of fuel assemblies at the core periphery. An important feature of this work is the capability for benchmarking of the detailed neutron and gamma spectra. The benchmarks were elaborated by the efforts of specialists from the Czech Republic, Russia, Germany, Bulgaria, Hungary, Ukraine, Spain and USA.
Benchmarking
Comparative analyses [6,7] of many calculations of neutron flux with energy above 0.5 MeV, based on the above mock-ups, carried out with different calculation methods (discrete ordinate and Monte Carlo), reactor models and nuclear data libraries, carried out by different teams, showed satisfactory levels of agreement (within 12%). This neutron flux is of great importance as it is used for assessment of the radiation metal embrittlement according to the Russian standard (Regulations, 1989). But considerable discrepancies between the calculational discrete ordinate and Monte Carlo results were found for the thermal neutron flux (17%-49%) and gamma flux integrals (10%-35%). These discrepancies are expected to be caused by the well-known fact that the multigroup cross sections for thermal (E<0.414 eV) neutrons and gamma radiation do not describe the interaction with elements sufficiently well. The TORT code calculates neutron and gamma fluxes with both BGL and BUGLE96 libraries. Comparison of both results shows that the BGL results are more consistent with MCNP code calculations, especially for gamma fluxes. This point is important since the BUGLE96 library is applied rather often for VVER calculations. However, it has to be stressed that both libraries mentioned are not adequate enough for thermal neutron and gamma calculations.
The comparison of the results of the independent calculations [7] shows:
■ reasonable agreement, within 15%, for fast neutron fluxes that satisfies the 15% correspondence requirement of the Russian standard [8] and the 20% correspondence requirement of the USA standard [9]
■ the calculated fast neutron attenuation through the RPV is relatively somewhat lower compared to experimental values, but attenuation is mostly somewhat higher over the downcomer,
■ the neutron and gamma spectra show generally larger discrepancies between calculation and experiment than the integral values.
For example, neutron and gamma flux spectra on the barrel of the mock-up of VVER-1000, calculated by different laboratories from Bulgaria (INRNE), Czech Republic (UJV; Skoda), Germany (FANP; FZR), Hungary (AEKI), and Spain (SEA), are presented in Figures 1 and 2. Comparison of experimental and calculated neutron and gamma flux spectra is demonstrated in Figure 3 and 4.
The results of the conformity study [6,7] between the attenuation data of VVER mock-ups obtained from the LR0 reactor, and from the RPVs of operational VVER nuclear power plants show that the VVER mock-ups could be used as benchmarks for validating the calculations of NPP RPV irradiation conditions. Still, applying the VVER-1000 mock-up results must take into account the NPP RPV attenuation azimuth dependence, and the difference between the biological shieldings of the mock-up and the reactor [10].
Damage attenuation
One of the important considerations in the study of radiation embrittlement of reactor vessels is the rate at which the radiation damage, in terms of DPA (displacements per atom) attenuates through the wall of the reactor pressure vessel. It is reasonable to assume that the embrittlement caused by neutrons depends on the energy of the neutrons, that is, on the energy spectrum. The neutron spectrum varies with the depth of penetration into the wall. DPA calculations can be used to obtain the effective vessel wall fluence for use in embrittlement trend curves, as proposed in the American Standard ASTM-E900-02 (2007) [9]. Alternatively, the exponential attenuation formula exp(-0.24x), where x is the distance in inch units into the vessel wall from the inside surface, may be used [11]. The TORT code+BGL cross sections library calculation for flux (E>0.5 MeV) attenuates through both VVER-440 and VVER-1000 RPV more slowly than trend curves calculated using DPA or the exp(-0.24) formula (Figure 5). This means that the E>0.5 MeV flux calculation is the most conservative of the four approaches for irradiation damage evaluation. That is why the ASTM standard, which uses the DPA concept for the RPV damage attenuation, cannot be applied to the Russian formula [8] for evaluation of the reference ductile-to-brittle transient temperature shift.
On the basis of the neutron-gamma results from the VVER mock-ups, it is shown that the gamma contribution to the total (neutron and gamma) irradiation damage of the RPV metal is not significant for VVERs in terms of DPA [7]. The gamma DPA contribution does not exceed 4%. However it is necessary to take into account other possible effects like gamma-annealing, interaction of gamma radiation with free-migrating defects, gamma influence on the cascade evolution, etc.
Validation of RPV fluence estimates
The validation of the neutron fluence on the RPV is currently carried out by activity measurements of ex-vessel detectors of niobium, nickel, iron, titanium, and copper foils placed in the air cavity between the vessel outer surface and the biological shielding. Such measurements are carried out at NPPs in the Czech Republic, Finland, Russia, Hungary, Ukraine, and Bulgaria. Data from ex-vessel detectors are also used for making decisions concerning core loading strategy and plant life management.
Comparison between the calculated and measured data gives a basis for a fluence adjustment procedure for determining the ‘best estimated’ neutron fluence. Experimental data for ex-vessel detectors contain information for adjustment of NPP-specific parameters. It has been shown that the uncertainty of RPV neutron fluence evaluation could be significantly reduced by such an adjustment [12].
Another way to verify the neutron fluence evaluation is to measure 54Mn activities of scraps/templates taken out from inner surface of RPV. However, the relatively short half life (312 days) of the radionuclide 54Mn, which is a product of neutron irradiation of 54Fe, gives reliable information on the fluence for only the last three years of operation.
As early as 1980, scraping samples for dosimetry purposes were taken from the RPV inner surface (cladding) at Loviisa 1 in Finland to improve the fluence estimates for the VVER-440 RPV. Three irradiations of ex-vessel (cavity) detectors have also been carried out (unit 1: 1984-85, 1998-99; unit 2: 2002-03), and further scraping samples were taken from the RPV in both units in 1986 [13]. A special device designed and manufactured by Škoda JS [14] was used for the scraping of samples of the inner cladding material of the Dukovany 3 RPV in the Czech Republic. This approach for fluence verification was applied at Kozloduy 1 and 2, Bulgaria [15]. Scraps were taken out from the inner surface of the unit 2 RPV (1992), and scraps (1995) and templates (1996) were taken out from the inner surface of the unit 1 RPV. The data of the activity of the templates from the VVER-440 reactor vessel of unit 1 were combined with the data of the detectors irradiated during the 18th cycle on the outer surface of the vessel. This combination of data was useful for the verification of neutron fluence attenuation through the VVER-440 vessel for reactor core loading with dummy cassettes [16]. The samples (templates) were also taken out from the inner RPV surface of several VVER-440 reactors in Russia for pressure vessel metal mechanical testing. Measured values of sample activity were used for pressure vessel dosimetry in combination with data from the abovementioned VVER-440 mock-up measurements [17]. All the authors reported that the 54Mn activity results of the scraps and templates were consistent (within 10%) with the calculated values.
Validation of surveillance specimens
Surveillance specimen testing is important for the precise and reliable evaluation of RPV embrittlement and RPV lifetime assessment. That is why a validation of the neutron fluence on the surveillance specimens of NPP is very important.
Neutron field parameters in the Kola NPP VVER-440 surveillance position (evaluated with Russian dosimeters by the Kurchatov Institute) have been demonstrated to have good agreement with the AMES Common Reference Dosimeter monitor sets (types ACORD1 and ACORD2) developed by NRG [5]. Differences between neutron fluxes with energy E>0.5 MeV from different detectors do not generally exceed 10%. The exception is data from niobium detectors, which is different by 20%. That disparity could be caused by insufficient accuracy of nuclear data for the reaction 93Nb(n,n’)93mNb, or by inaccuracy of the calculated spectral distribution of the neutron flux.
An upgraded neutron dosimetry procedure for VVER-440 surveillance specimens has been developed by the Kurchatov Institute. This procedure is based on measurements of the 54Mn activity of each of the surveillance specimens and neutron field computations. In contrast to earlier procedures, this new method correctly takes into account all pressure vessel internals, the influence of core pattern on the neutron field in surveillance specimen channels, and the dependence of spectral index SI0.5/3.0 on the axial coordinates of surveillance specimens [18].
These improvements can significantly change the interpretation of the surveillance specimen mechanical testing results and trend curves because of the dependence of irradiation embrittlement temperature shift on neutron fluence [19]. Those effects are why the results of the surveillance programs for all of the ‘old’ VVER-1000 have to be updated using the improved surveillance dosimetry.
Neutron fluence spectrum calculations were performed for the reactor pressure vessel of a VVER-1000, applying the TRAMO Monte Carlo code [20]. Activities measured earlier in Balakovo 1 by fluence monitors, placed in special Charpy surveillance containers, were compared to TRAMO results. The average deviation from the measurements was about 5%. Good agreement was demonstrated between the fluence spectra near the RPV inner side at the height of the core beltline to the spectra at the Charpy probe positions on top of the radial reflector.
An adjustment procedure for determination of best-estimate neutron fluence on the RPV has been developed for VVER-440 and VVER-1000 types of reactors [12, 21]. The adjustment has been based on uncertainty data for inelastic, elastic and absorption cross-sections of iron in the reactor steel construction; chromium inelastic, elastic and absorption cross sections; hydrogen and oxygen elastic cross sections; neutron source spectrum; neutron source spatial distribution; density of steel construction including baffle, barrel, cladding, and RPV; moderator density, i.e. water density in the gap between the baffle and the barrel, and in the downcomer; and RPV inner radius. The discrepancies between the calculated and measured values of NPP ex-vessel detectors responses from the reactions 54Fe(n,p)54Mn, 63Cu(n,α)60Co and 93Nb(n,n’)93mNb have been used for parameter adjustment. Only ex-vessel detector experimental data are appropriate for adjustment of NPP-specific parameters; but they can be useful. For instance, the adjustment applied for one-dimensional VVER models using the data for discrepancies between calculated and measured activity values of detectors irradiated at Units 1 and 5 of Kozloduy NPP has significantly reduced the neutron fluence uncertainty by a factor of about 1.5 and 2, respectively.
Retrospective dosimetry
Retrospective dosimetry uses structural materials in reactors that were not originally intended for dosimetry purposes. In principle the neutron fluence can be derived from the activity induced in a few milligrams of reactor material that have been obtained by scraping, drilling or nibbling from the location of interest. Material from the RPV in particular can be analysed in this way. Retrospective dosimetry can be applied to both operating facilities (with a ‘biopsy’) and shut-down facilities (with an ‘autopsy’). It has become a widespread and useful technique.
The retrospective dosimetry method was successfully used in the Dukovany NPP Unit 3 VVER 440 RPV after the 18th cycle. The calculated activities were compared with the measurements of the weld material, and in the cavity as well. The calculated neutron fluence in the weld was 2% higher than that adjusted from the weld material measurement [22].
The possibility to use the activation reaction 93Nb(n,n’)93mNb in RPV cladding material for retrospective assessment of neutron fluence from scraps are taken out of the RPV inner surface has been demonstrated [23-4]. In this case, the accuracy of measured activity that can be achieved is better than 5%; although there are some disadvantages. The presence of sufficient quantities of niobium in the material is the limiting factor for this method. The method also requires extra equipment (ICP-MS system or nuclear reactor). Researchers must also take into account the accumulation of 93mNb as a result of irradiation of Mo and the decay of the accumulated 93Mo. This effect can be significant after a long cooling time of irradiated samples. Additional efforts are needed to evaluate the impact of Nb activity arising from the irradiation of Mo present in the steel.
Diamond is currently used as a temperature monitor. For this purpose, diamond detectors are located inside the VVER-1000 surveillance assemblies. The use of diamond detectors for neutron fluence measurements has been proposed by RRC Kurchatov Institute, and is now under discussion [25]. Neutron irradiation causes extension of the diamond crystal lattice; this extension can be measured by x-ray diffractometry with high precision. In other words, the diamond can capture the neutron fluence over a long period of irradiation, up to the whole reactor lifetime. The detector could be calibrated appropriately for determination of neutron fluence with energy above 0.5 MeV. This type of detector can be used as an integrating device for determination of the neutron fluence accumulated during a long period of irradiation, up to the whole reactor lifetime. Diamond detectors have been calibrated using fluence values (above 0.5 MeV) that have been obtained by the manganese method in surveillance containers of 13 VVER-1000 units (irradiation time from 315 days to 8.8 years).
Improvements
Reactor dosimetry is an essential instrument for proper assessment of the effect of irradiation on reactor internals and RPV damage, and therefore for planning ways for improving both operation and plant life extension. Although reactor dosimetry provides a good enough description of the neutron field parameters through calculations and measurements, improvements of calculational tools will reduce the neutron fluence uncertainty.
The main difficulties in neutron fluence determination are indirectly related to the surveillance methodology:
■ no direct measurements on RPV and internals
■ shortcomings of surveillance assemblies’ design and location
■ relatively short half life (312 days) of the radionuclide 54Mn available in the metal
■ construction design uncertainties.
A view of possible means of improvement of the surveillance methodology as a part of the NPP lifetime justification is given below.
In terms of calculation, the main improvement that is necessary is in the presentation of neutron and gamma interactions with nuclei by appropriate cross sections. In particular:
■ new evaluations are needed for all cross-sections which have any importance in neutron and gamma transport and dosimetry reactions
■ the existing multigroup neutron-gamma cross-section libraries have to be improved in the region of thermal neutrons
■ the multigroup neutron-gamma cross-section libraries need to be supplemented with full covariance matrices, which are necessary for adjustment calculations
■ appropriate problem-oriented cross sections for the region of VVER-1000/320 surveillance specimens have to be generated for a more adequate description of the neutron transport in this region.
VVER-1000 surveillance specimen dosimetry should be improved and validated. Such work includes appropriate experiments on the LR-0 reactor, using different experimental techniques and applying different calculation methods. The same VVER-1000 mock-up can be used as a base for these new experiments. Experiments with modified surveillance assemblies on power reactors should be continued as well.
Thermal neutron dosimetry should be included in a list of priority problems, taking into account the role of thermal neutrons in materials degradation and activity accumulation. Thermal neutron dosimetry in VVER concrete biological shielding is also important for solving decommissioning problems, assessment of accumulated activity, and for optimization of design and placement of control system detection devices (ionization chambers) used in the reactor start-up. For this research, creation of a concrete shielding benchmark using the LR-0 VVER-1000 mock-up is a necessary task.
Niobium retrospective dosimetry of RPV and surveillance specimens seems to be promising, but its reliability and accuracy must be assessed. The scope of this assessment includes methods of niobium extraction from different materials, activity counting, evaluation of the molybdenum background, reliable niobium detector manufacturing, as well as the calculational determination of activity, etc. have to be verified. This niobium method could be applied to the cladding of VVER-440 and VVER-1000 RPV, and for the stainless steel capsules of surveillance specimens containing sufficient amounts of niobium. A standard methodology should be developed and validated by the collaboration of different laboratories in order to achieve a harmonized approach.
The diamond dosimetry method currently in development should include proper calibration of diamond detectors, for example by comparison with activation detectors over a long period, with well-documented irradiation and temperature data (or for relatively short irradiation at a high level of neutron flux). It should be pointed out that the diamond lattice extension also depends on the irradiation temperature, but recent experiments [25] showed that the VVER-1000 surveillance specimens’ temperature is practically stable at 300°C in all surveillance containers. The diamond detector has the advantage of being capable of summarizing the neutron fluence up to high fluence values corresponding to the NPP lifetime.
The validation of the neutron fluence on the RPV by activity measurements of ex-vessel detectors has to be a permanent task. These detectors are placed in the air cavity between the vessel outer surface and the biological shielding. Comparisons between calculations and measured data will give confidence in the calculation of neutron fluence data.
The VVER-1000 surveillance specimens’ dosimetry should be improved by a combined methodology that includes manganese, niobium and diamond dosimetry and precise calculation methods. The methodology should include the determination of the orientation of surveillance assemblies and surveillance specimens by comparison of the measured and calculated azimuth distributions of 54Mn activity in parts of the surveillance assembly and/or surveillance specimens and container material. The calculation of azimuth distributions must be carried out on the basis of detailed local power distributions in the fuel assemblies, which are the nearest ones to the surveillance assembly under consideration. Improved manganese dosimetry should be based on the 54Mn activity measurements in the surveillance specimen material and calculations taking into account the cycle-by-cycle history of detailed local power distribution in the upper parts of the fuel assemblies. Niobium dosimetry should be based on the measurements of 93mNb activity in the surveillance specimen and/or container material, and appropriate calculations taking into account the real orientation of surveillance assembly and specimens. Diamond dosimetry could be considered independent of the retrospective Nb method, overcoming the necessity to perform extraction of the Nb, which requires the corresponding equipment. Existing VVER-1000 surveillance data should be updated using the improved dosimetry results.
The adjustment procedure for VVER RPV dosimetry could be extended and improved by including additional and/or updated information. Further adjustment development should be focused on the RPV fluence evaluation in the direction of maximum irradiation. The lack of a reliable covariance matrix for cross section data is the most restrictive factor for further development. It also has to be noted that to increase the effectiveness of the adjustment approach, a reduction of the cross-section uncertainties is necessary. Although reducing cross-section uncertainties will have the greatest impact on the response uncertainty, this work has to be done as a separate task at specially designated facilities.
The influence of gamma irradiation on the RPV (including damage, annealing, interactions with free migrating defects, influence on the cascade evolution, etc) has to be evaluated properly after improving the neutron-gamma data.
Researchers should apply the retrospective dosimetry technique together with material testing of trepans that have been taken out from the RPVs of the VVER-440 reactors of Griefswald NPP, Germany after its shutdown. Such testing should give more definitive data regarding the neutron fluence attenuation and its relation to the radiation damage through the RPV wall. The possibility of irradiation of samples in the surveillance containers located at the inner surface of the Temelin VVER-1000 RPV has been noted. Such an irradiation programme could give more extensive and reliable materials embrittlement data for future investigations. The dosimetry and especially the retrospective dosimetry of the RPV internals and claddings deserves detailed consideration.
International cooperation in retrospective dosimetry has helped solve the important and difficult problems encountered during the development of RPV neutron fluence assessment methodology. Much work has been accomplished under the EC and IAEA. In particular, many meetings organized within the European Working Group on Reactor Dosimetry (EWGRD) have provided a good basis for analysis and evaluation of results. EWGRD started around 1960 under the sponsorship of Euratom. In 1993, the Working Group on Reactor Dosimetry on VVERs joined the EWGRD, and established the reactor dosimetry community in Europe. The goal of the EWGRD is to provide a platform for direct exchange of experience and know-how in reactor dosimetry and related programmes. In May 2011, the 14th International Symposium on Reactor Dosimetry is scheduled to take place at the Omni Mount Washington Resort, Bretton Woods, New Hampshire (USA). This Symposium is jointly sponsored by ASTM International and the European Working Group on Reactor Dosimetry. It is organized by ASTM Committee E10 on Nuclear Technology and Applications.
The existing collaboration should be continued as a basis for continuous improvement and quality assurance of the methodology.
Author Info:
Krassimira Ilieva and Sergey Belousov, Institute for Nuclear Research and Nuclear Energy of Bulgarian Academy of Sciences, 72 Tsarigradsko shosse Blvd, 1784 Sofia, Bulgaria;
Antonio Ballesteros, European Commission – Joint Research Centre, Institute for Energy, Unit “Safety of Present Nuclear Reactors”, P.O. Box 2 – 1755 ZG Petten, The Netherlands;
Sergey Zaritsky, Russian Research Centre Kurchatov Institute, Kurchatov sqr. 1, Moscow 123182, Russia.
