Radiation Protection Dosimetry (2012), Vol. 151, No. 2, pp. 354 –364 Advance Access publication 17 February 2012
doi:10.1093/rpd/ncr487
NEUTRON FIELD MEASUREMENTS OF THE CRNA OB26 IRRADIATOR USING A BONNER SPHERE SPECTROMETER FOR RADIATION PROTECTION PURPOSES H. Mazrou1,* and M. Allab2 1 Centre de Recherche Nucle´aire d’Alger (CRNA), 02 Boulevard Frantz—Fanon, B.P. 399, Alger, RP 16000, Alge´rie 2 Laboratoire SNIRM, Faculte´ de Physique, Universite´ des Sciences et de la Technologie de Houari-Boumediene (USTHB), 16111 Alger, Alge´rie *Corresponding author:
[email protected]
The present work deals with the Bonner sphere spectrometer (BSS) measurements performed, to support the authors’ MonteCarlo calculations, to estimate accurately the main characteristics of the neutron field of the 241Am–Be-based OB26 irradiator acquired for radiation protection purposes by the Nuclear Research Centre of Algiers. The measurements were performed at a reference irradiation position selected at 150 cm from the geometrical centre of the neutron source. The spectrometric system in use is based on a central spherical 3He thermal neutron proportional counter. The response matrix of the present spectrometer has been taken to be similar to the original Physikalisch-Technische Bundesanstalt (PTB) (Braunschweig, Germany) BSS’s response matrix, with a five bins per decade energy group structure, as there is no significant difference in the BSS’s physical characteristics. Thereafter, the authors’ BSS measurements were used together with MCNP5 results to unfold the neutron spectrum by means of MAXED and GRAVEL computer codes from the U.M.G. 3.3 package, developed at PTB. Besides, sensitivity analysis has been performed to test the consistency of the unfolding procedure. It reveals that no significant discrepancy was observed in the total neutron fluence and total ambient dose values following the perturbation of some pertinent unfolding parameters except for the case where a 10 bins energy structure was assumed for the guess spectrum. In this latter case, a 5 % difference was observed in the ambient dose equivalent compared with the reference case. Finally, a comparative study performed between different counting systems together with MCNP5 and predictive formulas results shows that they were globally satisfactory, highlighting thereby the relevance of the unfolding procedure and the reliability of the obtained results.
INTRODUCTION Traditionally, the characterisation of neutron fields always deals with neutron spectrometry. Resorting such a technique is also motivated by the valuable dosimetric information that can be extracted from the spectrum, in particular, when applications are mostly dedicated to radiation protection purposes. It is worth noting that none of the existing neutron dosimetric instruments can give accurate information, due to their respective shortcomings, over the wide energy range covered by neutrons encountered in most radiation protection applications. Nowadays, the development of adequate spectrometry instruments to cover a wide range of applications dealing with radiation protection constitutes a challenging program of research and development planned in specialised laboratories for years at often high costs. The systems of neutrons spectrometry available in these laboratories are each other different from each other, according to their final utilisation and subsequently to the expected accuracy. An irradiation system, of the OB26 type(1), has been acquired by the Nuclear Research Centre of
Algiers (CRNA) to provide a neutron beam for radiation protection purposes. This system consists of a 241 Am –Be radionuclide source of 185 GBq (5 Ci) activity inside a cylindrical steel-enveloped polyethylene container with a radially positioned beam channel. Because of its composition, filled with hydrogenous material, one could expect large modifications in the primary neutron beam, by neutron scattering on different materials present inside the irradiation facility, compared with a free-in-air situation. Thus, one of the suitable ways to characterise the neutron field of such a special delivered setup, over an energy range covering almost 10 decades, is to use a Bonner sphere spectrometer (BSS). This system was definitely retained, at CRNA, as a reference neutron measuring tool for radiation protections applications. It is based on a central spherical 3 He thermal neutron proportional counter, type SP9, and was acquired from Centronic Ltd, UK(2). In addition, the BSS measurements performed during this work will serve to support the authors’ previous extensive Monte-Carlo (MC) calculations(3, 4)
# The Author 2012. Published by Oxford University Press. All rights reserved. For Permissions, please email:
[email protected]
Downloaded from http://rpd.oxfordjournals.org/ by guest on August 21, 2012
Received October 6 2011, revised December 24 2011, accepted December 28 2011
NEUTRON FIELD MEASUREMENTS USING A BSS
BSS NEUTRON FIELD MEASUREMENTS Description of the CRNA BSS The 241Am –Be neutron spectrum has been measured using the authors’ lately acquired BSS based on a spherical 3He proportional counter(2). The CRNA Bonner sphere set consists of 12 available moderating polyethylene spheres of diameters ranging from 3 inch (7.62 cm) to 12 inch (30.48 cm). The mean polyethylene density was estimated to be (0.949+0.008) g cm – 3. In this work, configuration of five spheres of 3, 5, 8, 10 and 12 inches has been used in addition to the bare detector with and without cadmium. For similar spectra, such configuration is supposed to be sufficient for calculating integral quantities such as fluence rates or dose equivalent rates with reasonable accuracy for radiation protection purposes(8, 9). The authors consider that the main features of this spectrum can be sufficiently highlighted, firstly, for the thermal component (,0.5 eV) by the use of the bare detector with and without cadmium and secondly, for the peak (fast component) by the use of the sphere of 8 inch which is important because
the maximum response(10) is usually obtained with this sphere for this type of spectrum (ISO 8529(11) 241 Am –Be spectrum). The remaining spheres (3, 5, 10 and 12 inches) are used, for completeness, to define both sides of the curve around the peak with sufficient accuracy. In addition, it has been proved that measurements with a more sophisticated BSS system which uses 12 moderated spheres and unfolding for 53 energies (five energy bins per decade), to characterise similar spectra, did not introduce a difference by no .8 %(8). The Bonner sphere measurements The neutron detection is based on 3He (n, p) t reaction (Q¼ þ765 keV) with a cross section for thermal neutrons of 5321 barns. The full energy peaks at about channel 765, in the multi-channel analyzer (MCA), which corresponds to the reaction energy of 765 keV as shown in Figure 1. The discrimination level was placed at channel 153, after submitting the bare detector to pure gamma sources of 137Cs and 60 Co, respectively. This threshold allows a separation of the neutron-induced events from noise and gamma rays-induced events. The detector pulse height spectra were registered using Nuclear Instruments Modules electronics acquired from CANBERRA (2006 preamplifier, 2022 amplifier and MULTIPORT-II MCA). A high voltage of þ800 V has been applied to the 3He central detector. The counting time was different depending on the sphere used, but it was long enough (300–600s) to maintain the statistical uncertainty below 1 % for all spheres, except for the bare detector covered with cadmium where it needed 3600s irradiation time to keep the statistical uncertainty around 2 %. Corrections for dead-time were not necessary as the counting levels were around 100 counts s21 (,0.1 %). Each polyethylene sphere together with the central 3 He detector has been irradiated sequentially six times, at the irradiation position located at 150 cm source-detector distance (SDD). An example of the irradiation configuration in the secondary standard dosimetry laboratory (SSDL) bunker room is illustrated in Figure 2. The mean counting rates obtained for each sphere are reported in Table 1. In this table, one can see that the sphere diameter equal to 0 inch refers to the bare detector and the one equal to 0 inchþCd refers to the bare detector covered with cadmium. The obtained results show, as expected, that the maximum count rate is obtained for the sphere diameter of 8 inch as stated previously(10), whereas, the lowest count rate is obtained for the irradiation configuration of 0 inchþCd because of the low sensitivity of the 3He detector for epithermal and fast neutrons.
355
Downloaded from http://rpd.oxfordjournals.org/ by guest on August 21, 2012
to estimate accurately the neutron field delivered by the OB26 irradiator, at a reference irradiation position, set at 150 cm from the geometrical centre of the neutron source. However, using such a spectrometric approach will need to resort additionally to specific computer programs to unfold neutron spectra from the measured counts through a judicious unfolding procedure. In this case, the consistency of the results often depends upon prior information on the neutron spectrum shape and on the skilfulness of the user, as it has been illustrated in a previous paper(5). Hence, the ultimate goal of the present study is to develop and standardise a whole procedure based on BSS measurements, MC calculations and unfolding computer codes to characterise complex neutron/ gamma fields at any workplace, in particular, around the CRNA neutron OB26 irradiation facility. From this spectrum, one can deduce all the dosimetric quantities required for radiation protection applications. The authors have thus studied, in this work, the appropriateness of using two available unfolding codes, based on different unfolding methods, namely MAXED and GRAVEL, from the U.M.G. 3.3 package developed at PTB (Physikalisch-Technische Bundesanstalt) and made freely available through NEA data Bank(6), to determine the 241Am– Be neutron spectrum from the BSS measured data. The resulting spectra were thereafter compared with those obtained by an MC approach using the MCNP5(7) computer code.
H. MAZROU AND M. ALLAB
UNFOLDING PROCEDURE AND CODES
set by the user) following the next expression:
Neutron spectrum unfolding problem The deduction of the neutron spectrum from the Bonner sphere measurements counts is one of the typical problems encountered in multisphere spectrometry. This can be solved by applying unfolding procedures. Mathematical equations describing such systems are formally known as Fredholm integrals of the first kind given by(12): ð
_ dE Mi + 1i ¼ Ri ðEÞFðEÞ
ði ¼ 1; . . . ; nd Þ
x2 ¼
X 12 i
i
ð2Þ
s2i
Equation (1) has no unique solution since a finite number of discrete measurements, nd, cannot define _ a continuous function FðEÞ: It can however be approximated by the discrete Equation (3) where _ E is the fluence rate in energy group j extending F j from energy Ej to Ejþ1 (DEj ) and Ri represents Rij averaged over j groups:
ð1Þ Mi + 1 i ¼
nE X
_ E ðEj Þ DEj Rij ðEj Þ F j
ð3Þ
j¼1
where Mi is the sphere i reading obtained mathematically by folding the response function Ri (E) of this _ sphere with the spectral fluence rate FðEÞ, both expressed as functions of the energy E, whereas 1i is the difference between the predicted and the measured fluence values for the sphere i. 1i is related to the estimated standard uncertainty si via the chi-square statistical parameter x 2 (to be
Because the number of energy groups nE is usually greater, normally more than 50, than the number of Bonner spheres nd used, typically of the order 10, the derivation of neutron energy spectrum in such a case may provide a poor solution. These difficulties are therefore overcome by applying refined unfolding procedures developed in appropriate
356
Downloaded from http://rpd.oxfordjournals.org/ by guest on August 21, 2012
Figure 1. Example of a pulse height spectrum obtained for an irradiation configuration using a 3He central detector with a sphere of 8 inch diameter.
NEUTRON FIELD MEASUREMENTS USING A BSS
Table 1. BSS count rates for the CRNA OB26 neutron irradiator at 150 cm irradiation position. Sphere diameters (inch) 0 0þCd 3 5 8 10 12 a
-
Count rates (counts s – 1) 25.5+0.2 1.1+0.0a 35.5+0.3 89.5+0.3 116.1+0.5 100.5+0.7 80.7+0.7
from spectra approaching the solution, choose _ DEF _ i ; which is the closest as possible to F the F i among all feasible solutions which maximise the (6, 13) : entropy S defined by
S¼
X i
" _ i ln F
_i F _ DEF F i
!
_i _ DEF F þF i
# ð4Þ
_ i is the determined fluence rate in group where F _ DEF is its corresponding value in energy i and F i the discretised default spectrum.
0.0 refers to 0.02.
unfolding computer codes based on different mathematical principles and used successfully to solve such complex unfolding problems. The two unfolding codes used in this work, namely MAXED and GRAVEL, are briefly described hereafter emphasising on the assumptions assumed during the calculations.
Unfolding using maxed code Unfolding method is based on the following steps: _ DEF to - choose the best estimate of the spectrum F i start the iterative calculations;
Running MAXED code always requires creating, beforehand, input files including information on: -
the response functions of the spectrometric system to be used; the collected spectrometric measurements and _ DEF ( prior information) to the default spectrum F i be used to initiate the iterative calculation.
During the optimisation process, the search for a maximum of S is performed using a special program called simulated annealing. For more details, the reader is referred to relevant references available in the literature(6, 12 – 14).
357
Downloaded from http://rpd.oxfordjournals.org/ by guest on August 21, 2012
Figure 2. General view of the OB26 neutron calibrator irradiating an 8 inch sphere diameter inside the Algerian secondary standard dosimetry laboratory (SSDL) calibration room.
H. MAZROU AND M. ALLAB
Previously obtained(15) energy response functions of a few groups are introduced in MAXED in order to unfold the neutron spectrum (see hereafter section ‘Matrix response’). The MCNP5 spectrum calculated with 5 and 10 bins energy structure [see section ‘The default spectrum ( prior information)’] has been used as default spectrum during the unfolding procedure. The x 2-value was set equal to 0.1 after several runs. Unfolding using gravel code
kþ1
_ is the value of the fluence rate in energy where F j group j at the (k þ 1)th iteration; Mi is the count rate measured by the detector i; Cik is the count rate of the detector i calculated at iteration k: Cik ¼
ng X
k
_ Þ Rij expðln F j
ð6Þ
i¼1
Wijk is the relative contribution of energy group j in the detector response i during the run k: Wijk
¼
_k Rij F j Cik
ð7Þ
The solution is obtained by minimising the statistical _ i Þ; i ¼ 1; . . . ; nd : To find parameter x 2 versus ðln F 2 the minimum of x , the code uses a nonlinear leastsquare method. For more details, the reader is referred to relevant references available in the literature(6, 16, 17). The x 2-value was fixed equal to 0.2 after several runs. The default spectra used in GRAVEL code have been taken from MCNP5 calculations described in the following section ‘The default spectrum ( prior information)’. Matrix response Most of the neutron spectrometers currently in use are generally well characterised and their energy responses are relatively well known. So, to characterise neutron beams at specific sites, it is often referred to these response functions provided that the average polyethylene density of the used spectrometer and the gas pressure of the detector are the same as those the response matrix of which are available in the literature. If not, these responses should be
The default spectrum ( prior information) As stated in sections ‘Unfolding using maxed code’ and ‘Unfolding using gravel code’, the MCNP5 neutron spectra, with two different energy structures (5 and 10 bins), have been used as default spectra for both unfolding codes GRAVEL and MAXED. These spectra were obtained from the OB26 neutron irradiator at a SDD of 150 cm selected in the authors’ previous MCNP5 calculations(4). In the present work, additional MC calculations have been performed using a new bins energy structures, consisting respectively of 5 and 10 log-equidistant bins per decade to cover the whole interval of energy of interest (1023 eV to nearly 20 MeV), the earlier bin structure being similar to that adopted for the matrix response of the BSS(15). The MC calculations were performed assuming an ISO 8529/1(11) primary spectrum to represent the 241 Am –Be neutron source one. The materials used during MCNP5 modeling consist, as shown in Figure 3, mainly of: 241Am–Be neutron source, polyethylene, air, lead, cadmium and wall concrete of the irradiation room. The stainlesssteel of 304 type was used for both X14 capsule and the OB/26 container. The elemental composition (weight percentage) and densities of these materials, as considered in MCNP5, were given in detail in the authors’ previous work(4). The MCNP5 fluence results obtained for the actual irradiation configuration ( previously labelled Conf_4)(4) are represented in Figure 4 using two (5 and 10 bins) energy structures. The fluence distribution is given per lethargy unit [i.e. per logarithmic energy interval Dln (E)] as defined by ISO 8529(11). In this work, thermal neutrons are the particles below the ‘cadmium cut-off energy’, i.e. 0.5 eV, the
358
Downloaded from http://rpd.oxfordjournals.org/ by guest on August 21, 2012
At iteration k, the fit procedure is performed following Equations (5 –7)(6) given hereafter: "Pn # d k k kþ1 k i¼1 Wij lnðMi =Ci Þ _ _ Pnd Fj ¼ Fj exp ð5Þ k i¼1 Wij
corrected by an appropriate correction factor to account for the noted differences. In this work, the density was averaged over the all the used spheres. It was found equal to 0.949 g cm – 3, a value nearly equal (0.4 %) to that (r¼0.946 g cm – 3) for which response functions were given in the literature by Alevra and Plostinaru(15). As regards the effects of the variation of the gas (3He) pressure in the SP9 detector on the response functions, it has been stated in the literature that a reduction in gas filling by 10 % produces a decrease of BSS responses by 5 %(15). In this study, the response functions taken from the literature(15) were those given for a nominal gas pressure (200 kPa) of the SP9 detector. Therefore, in this work, the response functions reported in this reference with an energy distribution of five bins per decade were adpoted. The response functions with 10 bins energy structure have been obtained from the five bins energy structure using a ‘cubic spline’ interpolation method.
NEUTRON FIELD MEASUREMENTS USING A BSS
irradiator shielding and to the multiple neutron scattering occurring with concrete walls of the SSDL irradiation room(4). ANALYSIS OF BSS RESULTS Unfolded neutron fluence spectra compared with MCNP5 ones
Figure 4. MCNP5 fine energy (05 and 10 bins) spectra for 241 Am– Be-based-neutron irradiator for real SSDL configuration at the selected distance SDD¼150 cm.
epithermal ones were defined in the energy range between 0.5 eV to 10 keV and the fast neutrons were defined to be above 10 keV. Figure 4 reveals clearly, as expected, the presence of a thermal part in the neutron spectrum due to the massive presence of hydrogenous material in the composition of the OB26
In Figure 5, are given the energy spectrum for the SSDL/CRNA neutron field calculated by MCNP5 together with BSS-based spectra obtained by MAXED and GRAVEL unfolding codes. These latter were run assuming a guess spectrum described by the five bins MCNP5 calculations given in the preceding section ‘The default spectrum ( prior information)’ (Figure 4). The deduced dosimetric quantities were calculated using the recommended ICRP-74(18) fluence-to-dose equivalent conversion coefficients for the MCNP5 spectrum and for the MAXED and GRAVEL ones. In Table 2, important dosimetric parameters are reported, namely the total neutron fluence F and the total ambient dose equivalent H*(10). The values of these integral quantities extracted from both unfolding codes (MAXED and GRAVEL) are very consistent with each other
359
Downloaded from http://rpd.oxfordjournals.org/ by guest on August 21, 2012
Figure 3. Schematic view of 241Am– Be neutron calibrator OB26(1). (dimensions are in mm).
360
292.9 301.7 301.3 391(11)
h*(10) (pSv cm2)
79.7+3.0 64.7+2.4 63.6+2.3 — 79.0+2.9 64.2+2.3 63.1+2.2 — *E, Energy is given in MeV. a 0.0 refers to values ,,0.1.
MCNP5 BSS (MAXED) BSS (GRAVEL) 241 Am– Be
12.7+0.5 9.2+0.3 8.9+0.3 —
4.9+0.2 2.7+0.1 3.0+0.1 —
58.0+2.1 48.1+1.8 47.5+1.7 —
75.6+2.8 60.0+2.2 59.4+2.1 —
0.5+0.1 0.4+0.1 0.4+0.1 —
0.2+0.0a 0.1+0.0a 0.1+0.0a —
Total E* 51027
51027 ,E 1022
E . 1022
Total
E 51027
51027 , E 1022
E . 1022
3.3 3.2 2.9 4.2(11)
E¯ (MeV) H*(10) (mSv h21) f (cm22 s21) Codes
_ _ ð10Þ ¼ 64:1 mSv h1 with ½F ¼ 59:7 cm2 s1 and H a difference below 2 %. This suggests that both codes, which are based on different mathematical approaches, converge to a single solution when a good a priori data are available. However, a discrepancy of 25 % is observed, for the two integral quantities of interest, when _ ¼ 75:6cm2 s1 comparing MCNP5 results ½F _ ð10Þ ¼ 79:7mSvh1 with both MAXED and and H GRAVEL ones. This discrepancy may be due, among others, to the significant uncertainty that affects the nominal source strength which has been given by the manufacturer around 10 %. The lower H*(10) values obtained for the unfolded spectra may also be due to the softer spectra in the fast region (additional fluence on the low energy side on the peak). On the contrary, the mean values of the energy and the fluence to ambient dose spectrum ðEÞ equivalent ½h ð10Þ conversion coefficients given by the three codes, also reported in Table 2, are very close to each other, with no significant differences. As uncertainties, the sum in quadrature of the statistical uncertainty (1 –2 %) and the response matrix uncertainty (3.5 %) were used for both unfolding computer codes (MAXED and GRAVEL) and an overall 10 % uncertainty for the source strength was added to 1 % statistical uncertainty for MCNP5 results. Consequently, the resulting uncertainties are around 4 % for the unfolding results and around 10 % for the MCNP5 ones. This latter uncertainty can be considered as a conservative value. Overall, these uncertainty calculations can be improved by using a more elaborate method like the one recommended
Downloaded from http://rpd.oxfordjournals.org/ by guest on August 21, 2012
Figure 5. Neutron spectra of the CRNA neutron calibrator obtained at an irradiation position (SDD¼150 cm) using two unfolding codes (MAXED and GRAVEL) compared with MCNP5 guess spectrum using five bins energy structure.
Table 2. Comparison between calculated (MCNP5) and measured (BSS) neutron fluence rate, ambient dose equivalent rate, mean energy and spectrum averaged fluence-todose equivalent conversion coefficient for the CRNA irradiator at 150 cm irradiation position.
H. MAZROU AND M. ALLAB
NEUTRON FIELD MEASUREMENTS USING A BSS
by ISO(19), referring to the Guide to the expression of Uncertainty in Measurements. This will be the next challenge of the dosimetry laboratory of the CRNA, in a near the future, to standardise its uncertainty calculations.
Consistency of the unfolding procedure To test the consistency of the unfolding results, the ratio rd between calculated counts Cd and measured ones Md for each sphere diameter has been estimated: rd ¼
ð8Þ
Situations where the calculated counts Cd are close to the true counts (measured) Md are described by rd values close to unity. The values of rd are illustrated in Figures 6 and 7 for unfolding codes MAXED and GRAVEL, respectively. The sphere label one is for the bare detector shielded with cadmium. The uncertainties include the statistical errors on measurements and the errors due to response matrix of the BSS taken from the literature(15). It is found that overall the average ratio is close to unity for both unfolding codes (for GRAVEL: rd ¼ 0:99946 + 0:41 % and for MAXED: rd ¼ 0:99984 + 0:34 %) highlighting the reliability of the unfolded results.
To test the sensitivity of the results deduced from the unfolding procedure to some important input parameters, the authors have quantified the effect of a perturbation of some of them, involved in the unfolding process used in both codes (MAXED and GRAVEL), on these results. The variations of physical and dosimetric quantities, namely the variation of the fluence, dF, and that of the ambient dose equivalent, dH*(10), are calculated in relation to a reference case. This later was defined, after several sensitivity tests performed on the codes to reach a convergent solution, with the following initial parameters: -
the initial MC spectrum calculated with an energy distribution of five bins per decade (see Figure 4); - a statistical factor x 2 ¼0.1 and - a reduction factor a¼0.85 (convergence parameter used during the optimisation process in the simulated annealing method); Variations stated successively to these parameters are summarised as follows, taking into account a maximum number of iterations set at 2000 (stopping criterion): -
The initial MC default spectrum test was first replaced by a so called ‘Educated guess spectrum’, defined as the spectrum issued from the first unfolding calculations with both MAXED and GRAVEL codes (reported in Figure 5), then by the spectrum from the MC calculations assuming
Figure 6. Ratios of the calculated readings (MAXED) to the measured readings (BSS), derived from neutron spectra shown in Figure 4.
361
Downloaded from http://rpd.oxfordjournals.org/ by guest on August 21, 2012
Cd Md
Sensitivity analysis of the unfolding procedure
H. MAZROU AND M. ALLAB
Table 3. Sensitivity analysis of fluence and ambient dose equivalent to few parameters on the unfolding procedure based on MAXED and GRAVEL codes. Parameters
MAXED f (cm22 s21) (df, %)
Reference case using a five bins energy structure Educated guess spectrum with five bins energy structure Initial MC guess spectrum with 10 bins energy structure x2 ¼1.0 x2 ¼0.01 a ¼0.95 (reduction factor)
GRAVEL
H*(10) (mSv h21) (dH*, %)
f (cm22 s21) (df, %)
H *(10) (mSv h21) (dH*, %)
60.0
64.7
59.4
63.6
60.3 (þ0.6)
65.5 (þ1.2)
60.3 (þ1.5)
65.4 (þ2.8)
59.0 (21.7)
61.5 (25.1)
58.7 (21.2)
61.1 (24.1)
59.7 (20.5) 60.7 (þ1.3) 59.0 (21.7)
63.8 (21.4) 66.2 (þ2.3) 63.5 (21.8)
59.5 (þ0.3) 59.6 (þ0.4) —
63.5 (20.1) 64.0 (þ0.6) —
an energy distribution of 10 bins per decade shown in Figure 4. - The statistical factor, x 2: as the process of calculation is iterative in both codes, the choice of this factor is essential for the rapid convergence of the solution. Consequently, two other values considered as reasonable (0.01 and 1.0) have been taken. - The a reduction factor: this factor, connected with the simulated annealing method, is used exclusively by the MAXED code to find the global optimum. This parameter defines the speed of convergence of the used method. If it is too large (near unity), the method does not converge
quickly, if it is too small, the method may not lead to the solution (final solution to be achieved). Then used properly, it offers the advantage of avoiding precisely that the solution will be trapped in a local optimum (intermediate solution that one hopes to avoid). The recommended value in the code MAXED is 0.85. Calculations have been made with a¼0.95. The fluence F and ambient dose equivalent H*(10) obtained in each calculation are summarised in Table 3, where are also reported the variations dF and dH*(10) are also reported in relation to the reference case.
362
Downloaded from http://rpd.oxfordjournals.org/ by guest on August 21, 2012
Figure 7. Ratios of the calculated readings (GRAVEL) to the measured readings (BSS), derived from neutron spectra shown in Figure 4.
NEUTRON FIELD MEASUREMENTS USING A BSS
Analysis of the results reveals the following remarks: -
Following the above remarks, a new unfolding procedure was then performed based on the initial educated guess spectrum (spectrum solution of MAXED and GRAVEL codes) with an energy distribution of 10 bins per decade. So, after new unfolding calculations, the mean integral quantities achieved by the two codes (MAXED and GRAVEL) are as follows: _ _ ð10Þ ¼ 63:3 mSv h1 : F ¼ 59:7 cm2 s1 ; H The results obtained in this case will now constitute the new reference one for the following section. COMPARATIVE STUDY OF THE AMBIENT DOSE EQUIVALENT RATES The various ambient dose equivalent rates determined by different approaches (experimental and theoretical or MC ones), for the OB26 irradiator neutron field at a reference SDD of 150 cm, have been compared. The ambient dose equivalent rates obtained by both area dosemeters and by predictive formulas have been taken from the authors’ previous work(4). The MC values have been obtained in this work (given in Table 2) using the nominal source emission rate value corrected to the date of measurements (1.08 107 +10 % s21). The ambient dose equivalent _ ð10Þ ¼ 63:3mSvh1 Þ obtained by the BSS is rate ðH taken from a preceding section and it represents the mean value of both MAXED and GRAVEL codes assuming an educated guess spectrum with an
Instrument Reference instrument Bonner sphere spectrometer Conventional area dosemeter(4) Berthold LB 6411 Studsvik 2222A Studsvik 2202D Computer code MCNP5 Predictive formulas(4) Mean+SD
Ratios
1.00 1.13 0.99 1.10 1.26 0.96 1.09+11.1 %
energy distribution of 10 bins per decade. The ratios of these ambient dose equivalent rates to that obtained by BSS, taken as the reference case, are reported in Table 4. The ratio values reveal an overall good agreement between the results obtained by conventional dosemeters (Berthold and Studsvik) and calculations (MCNP5 and predictive formulas) and those measured by BSS. The discrepancy does not exceed 25 % (MCNP5 case) due to reasons previously stated. The mean ratio over the overall results is 1.09+11 %. This shows the consistency of these results and can constitute an additional validation of the unfolding procedure. CONCLUSION In this study, a spectral characterisation of the neutron beam issuing from an OB26 irradiator using the multisphere spectrometer BSS acquired by CRNA has been carried out. The obtained BSS results were used to support computational MC results. The 241Am –Be neutron spectrum has been extracted from BSS measurements, through an unfolding procedure, at a 150 cm reference irradiation position from the geometrical centre of the neutron source, by means of two unfolding codes (MAXED and GRAVEL), widely used by many specialised laboratories. Following a sensitivity analysis performed on some selected parameters to test the consistency of the unfolding procedure, the results obtained (fluence and ambient dose equivalent rates) are very encouraging. They show, in particular, that the final solution, after multiple iterations, converges towards ‘true’ values. In addition, this suggests the relevance of the initial parameters chosen, in particular, the a priori data in both unfolding codes. A comparative analysis between different dosimetric instruments results together with MCNP5 and
363
Downloaded from http://rpd.oxfordjournals.org/ by guest on August 21, 2012
Overall, the sensitivity to deconvolution is considered low for both variables selected (fluence and ambient dose equivalent). The largest difference recorded in relation to the reference case is 5 % and it has been obtained, for the ambient dose equivalent quantity, when the initial guess spectrum presents an energy distribution of 10 bins per decade. This difference is mostly due to the values of the fluence to ambient dose equivalent conversion coefficients which are very sensitive to the energy structure adopted. For the remaining parameters including the educated guess spectrum F0, the statistical factor x 2 and the reduction factor a, sensitivity to unfolding procedure is even lower (less than +3 %). This suggests the relevance of the initial parameters chosen, in particular, the a priori data in both codes and the consistency of resulting solution. - The convergence was fast for the case of the educated guess spectrum and the resulting solution is identical for both codes deconvolution. This suggests that the authors are closer to the ‘real’ solution.
Table 4. Ratios between the BS measurements (unfolded with MAXED and GRAVEL using educated guess spectrum with a 10 bins energy structure) and conventional area dosemeters, MCNP5 results and predictive formulas.
H. MAZROU AND M. ALLAB
ACKNOWLEDGEMENTS One of the authors (H. Mazrou) wishes to express his heartfelt thanks for his colleagues from the research centres of CRNA and CRND for their valuable assistance, fruitful discussions and keen interest given to the present work and, in particular, to O. HAROUA (CRNA) for his kind assistance provided from the receipt till the final set-up of NIM electronic modules in the whole experimental setup. REFERENCES 1. STS, Steuerungstechnik & Strahlenschutz GmBH. Neutron Calibrator OB26/2: Technical Specifications. Braunschweig (1998). 2. CENTRONIC LIMITED. 3He Proportional Neutron Counter of SP9 Type: Technical Specifications (2007). 3. Mazrou, H., Sidahmed, T. and Allab, M. Monte-Carlo investigation of radiation beam quality of the CRNA irradiator for calibration purposes. Appl. Radiat. Isot. 68(10), 1915– 1921 (2010). 4. Mazrou, H., Sidahmed, T. and Allab, M. Neutron field characterization of the OB26 CRNA irradiator in view of its use for calibration purposes. Radiat. Prot. Dosim. 141(2), 114–126 (2010). 5. Alevra, A. V. and Thomas, D. J. Neutron spectrometry in mixed fields: multisphere spectrometers. Radiat. Prot. Dosim. 107(1– 3), 37–72 (2003). 6. Reginatto, M., Wiegel, B., Zimbal, A. and Langner, F. Analysis of data measured with spectrometers using unfolding techniques. NEA-1665/03 UMG 3.3 (2004). 7. MCNP5. A General Monte-Carlo N-particle Transport Code. Version 2.1. X-5 Monte Carlo Team. Report LAUR-03-1987. Los Alamos National Laboratory (2003).
8. Rimpler, A. Bonner sphere neutron spectrometry at spent fuel casks. Nucl. Instrum. Methods A 476, 468– 473 (2002). 9. Wang, Z., Hutchinson, J. D., Hertel, N. E., Burgett, E. and Howell, R. M. Study of a gold-foil-based multisphere neutron spectrometer. Radiat. Prot. Dosim. 128(3), 289– 293 (2008). 10. Mazrou, H., Idiri, Z., Sidahmed, T. and Allab, M. MCNP5 evaluation of a response matrix of a Bonner sphere spectrometer with a high efficiency 6LiI(Eu) detector from 0.01 eV to 20 MeV neutrons. J. Radioanal. Nucl. Chem. 284(2), 253– 263 (2010). 11. International Organization for Standardization. Reference neutron radiations, Part 1: characteristics and methods of production. ISO 8529-1. ISO (2001). 12. Matzke, M. Unfolding procedures. Radiat. Prot. Dosim. 107(1– 3), 155–174 (2003). 13. Reginatto, M., Goldhagen, P. and Neumann, S. Spectrum unfolding, sensitivity analysis and propagation of uncertainties with the maximum entropy deconvolution code MAXED. Nucl. Instrum. Methods A 476, 242 (2002). 14. Reginatto, M. and Zimbal, A. Bayesian and maximum entropy methods for fusion diagnostic measurements with compact neutron spectrumeters. Rev. Sci. Instrum. 79, 023505 (2008). 15. Alevra, A. V. and Plostinaru, V. D. Characterisation of the IPNE Bonner Sphere Spectrometry by comparison with the PTB System, PTB-6.42-00-1. PTB report, Braunschweig (2000). 16. McElroy, W. N., Berg, S., Crockett, T. and Hawkings, R. G. A computer-automated iterative method for neutron flux spectra determination by foil activation. U.S. Air Force Weapons Laboratory Report AFWLTR-67-41, Kirkland AFB, NM (1967). 17. El Messaoudi, M., Chouak, A., Lferde, M. and Cherkaoui, R. Performance of three different unfolding procedures connected to bonner sphere data. Radiat. Prot. Dosim. 108(3), 247–253 (2004). 18. International Commission on Radiological Protection. Conversion coefficients for the use in radiological protection against external radiation. ICRP Publication 74 (1996). 19. International Organization for Standardization. GUM: Guide to the Expression of Uncertainty in Measurements. ISO (1995). ISBN 92-67-10188-9. 20. Mendez-Villafane, R., Los Arcos Merino, J. M., Gallego Diaz, E. and Lorente Fillol, A. Determination of the emission rate of an Am-Be neutron source with a Bonner sphere spectrometer. Radiat. Meas. 45, 1271–1275 (2010). 21. Fernandez, F., Domingo, C., Amgarou, K., Castelo, J., Bouassoule, T., Garcia, M. J. and Luguera, E. Neutron measurements in a Varian 2100c linac facility using a Bonner sphere system based on passive gold activation detectors. Radiat. Prot. Dosim. 126(1–4) 361–365 (2007). 22. Alem-Bezoubiri, A., Mazrou, H. and Bezoubiri, F. Simulation par la me´thode Monte-Carlo de la production des photoneutrons autour d’un acce´le´rateur line´aire me´dicale. Editions Universitaires Europe´ennes (2011) ISBN 978-613-1-58404-6. 23. International Organization for Standardization. Reference radiation fields—simulated workplace neutron fields. Part 1: characteristics and methods of production. ISO 12789-1. ISO (2008).
364
Downloaded from http://rpd.oxfordjournals.org/ by guest on August 21, 2012
predictive formulas results with the BSS ones is globally satisfactory. This suggests that the unfolding procedure is appropriate and the obtained results are reliable. However, much more agreement with the BSS ambient dose equivalent rate can be obtained with MCNP5 results, in particular, when assuming a new value of the emission source of the OB26 irradiator. This work using a similar approach based on BSS measurements, as described recently in the literature(20), is in progress. Overall, the obtained experimental results coupled to MC ones, can provide a best estimate of the OB26 neutron spectrum over a large energy range covering 10 decades (i.e. from thermal to MeV region) at a reference SDD of 150 cm. The neutron field, of the CRNA OB26 irradiator, has now been fully characterised at a reference irradiation location (SDD¼150 cm). The neutron energy spectrum resulted is quite similar, in shape, to those obtained in the literature(21, 22) around medical linear accelerators. This suggests that this OB26 irradiator may be used to reproduce neutron fields at similar workplaces as recommended by the related ISO 12789(23).