+ All Categories
Home > Documents > Simulation methods for photoneutron-based active interrogation systems

Simulation methods for photoneutron-based active interrogation systems

Date post: 11-Sep-2016
Category:
Upload: michael-king
View: 218 times
Download: 1 times
Share this document with a friend
4
Simulation methods for photoneutron-based active interrogation systems Michael King n , Timothy Shaw, John Stevenson, Mashal Elsalim, Craig Brown, Cathie Condron, Tsahi Gozani Rapiscan Laboratories, Inc., 520 Almanor Avenue, Sunnyvale ,CA 94085, USA article info Available online 30 August 2010 Keywords: Active interrogation Photoneutron MCNPX Simulation abstract Modeling methods have been developed to accelerate the simulation of a photoneutron-based active interrogation system of nuclear materials. The proposed technique segments the simulation of a full system into several physical steps, representing functional approximations. Each approximation is carried-out separately, resulting in a major reduction in computational time and a significant improvement in tally statistics. Although more human effort is required to separate each step, the net time required to produce results is drastically reduced. In addition, the results of previous steps can be used as inputs to proceeding steps without the need for re-simulation. We show that for a photoneutron interrogation system, the final results are in good agreement with the full, single-step simulation and also with experimental results. & 2010 Elsevier B.V. All rights reserved. 1. Introduction Active interrogation techniques have been developed to address the current deficiencies in air and seafaring cargo container inspection for hidden nuclear materials. These non- intrusive inspection systems utilize highly penetrating probing radiation to induce fission and sensitive detectors to detect characteristic fission signatures. One approach that is capable of fulfilling the inspection needs is based on fast-neutron interroga- tion [1]. An intense neutron interrogation system, which utilizes photoneutrons generated from Bremsstrahlung photons [2], has the ability to penetrate dense cargo and induce fission. For optimal performance, the development of the final design must consider many issues including extrapolation of experimental data, system sensitivity versus source strength, number of detectors and their distribution, radiation shielding and cost, all of which require extensive simulation and modeling. In this paper, an effort is made to model and corroborate experimental data as well as to determine the detection efficiencies of a full-scale photoneutron inspection system [1]. The photoneutron-based interrogation system utilizes a 9 MV Bremsstrahlung source to generate photoneutrons from either a beryllium or a heavy water converter to inspect a cargo container for the presence of nuclear material. The simulation of the system is segmented into multiple, manageable steps, each utilizing only common desktop quad-core computers. We will show that these segments minimize computational time as well as demonstrate the underlying physical events of each step, which allows for efficient performance optimization of the system. 2. Computational approach The Monte Carlo N-particle eXtended (MCNPX) [3] simulation code is chosen for this task because it provides the user with the powerful capability to model radiation transport in all nuclear- related applications. The code encompasses an extensive nuclear cross-section database, validated physics models and a strong user community. Our simulation effort to segment each change in radiation particle involves multiple steps: approximating an electron linac with a Bremsstrahlung source, developing a volume photoneutron source and approximating the emission and energy spectrum of fission signatures. 3. MCNPX simplification methodology The simulation code, MCNPX, is able to perform complex, full-scale simulations of any radiation inspection system by following the transport processes of radiation particles. A photoneutron-based active interrogation system detects nucle- ar materials when induced fission signatures are identified by the detectors. Fig. 1 shows all the key radiation conversion steps that appear in the interrogation system. The system utilizes an electron accelerator to generate a Bremsstrahlung beam. The Bremsstrahlung photons then produce photoneutrons in a con- verter, which subsequently induce fission in the nuclear material (NM, such as 235 U, 238 U, etc.), if present, in the cargo. Finally, Contents lists available at ScienceDirect journal homepage: www.elsevier.com/locate/nima Nuclear Instruments and Methods in Physics Research A 0168-9002/$ - see front matter & 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.nima.2010.08.041 n Corresponding author. Tel.: + 1 408 961 9727; fax: + 1 408 727 8748. E-mail address: [email protected] (M. King). Nuclear Instruments and Methods in Physics Research A 652 (2011) 129–132
Transcript
Page 1: Simulation methods for photoneutron-based active interrogation systems

Nuclear Instruments and Methods in Physics Research A 652 (2011) 129–132

Contents lists available at ScienceDirect

Nuclear Instruments and Methods inPhysics Research A

0168-90

doi:10.1

n Corr

E-m

journal homepage: www.elsevier.com/locate/nima

Simulation methods for photoneutron-based active interrogation systems

Michael King n, Timothy Shaw, John Stevenson, Mashal Elsalim, Craig Brown,Cathie Condron, Tsahi Gozani

Rapiscan Laboratories, Inc., 520 Almanor Avenue, Sunnyvale ,CA 94085, USA

a r t i c l e i n f o

Available online 30 August 2010

Keywords:

Active interrogation

Photoneutron

MCNPX

Simulation

02/$ - see front matter & 2010 Elsevier B.V. A

016/j.nima.2010.08.041

esponding author. Tel.: +1 408 961 9727; fax

ail address: [email protected] (M.

a b s t r a c t

Modeling methods have been developed to accelerate the simulation of a photoneutron-based active

interrogation system of nuclear materials. The proposed technique segments the simulation of a full

system into several physical steps, representing functional approximations. Each approximation is

carried-out separately, resulting in a major reduction in computational time and a significant

improvement in tally statistics. Although more human effort is required to separate each step, the net

time required to produce results is drastically reduced. In addition, the results of previous steps can be

used as inputs to proceeding steps without the need for re-simulation. We show that for a

photoneutron interrogation system, the final results are in good agreement with the full, single-step

simulation and also with experimental results.

& 2010 Elsevier B.V. All rights reserved.

1. Introduction

Active interrogation techniques have been developed toaddress the current deficiencies in air and seafaring cargocontainer inspection for hidden nuclear materials. These non-intrusive inspection systems utilize highly penetrating probingradiation to induce fission and sensitive detectors to detectcharacteristic fission signatures. One approach that is capable offulfilling the inspection needs is based on fast-neutron interroga-tion [1]. An intense neutron interrogation system, which utilizesphotoneutrons generated from Bremsstrahlung photons [2], hasthe ability to penetrate dense cargo and induce fission. Foroptimal performance, the development of the final design mustconsider many issues including extrapolation of experimentaldata, system sensitivity versus source strength, number ofdetectors and their distribution, radiation shielding and cost, allof which require extensive simulation and modeling.

In this paper, an effort is made to model and corroborateexperimental data as well as to determine the detectionefficiencies of a full-scale photoneutron inspection system [1].The photoneutron-based interrogation system utilizes a 9 MVBremsstrahlung source to generate photoneutrons from either aberyllium or a heavy water converter to inspect a cargo containerfor the presence of nuclear material. The simulation of the systemis segmented into multiple, manageable steps, each utilizing onlycommon desktop quad-core computers. We will show that thesesegments minimize computational time as well as demonstrate

ll rights reserved.

: +1 408 727 8748.

King).

the underlying physical events of each step, which allows forefficient performance optimization of the system.

2. Computational approach

The Monte Carlo N-particle eXtended (MCNPX) [3] simulationcode is chosen for this task because it provides the user with thepowerful capability to model radiation transport in all nuclear-related applications. The code encompasses an extensive nuclearcross-section database, validated physics models and a stronguser community. Our simulation effort to segment each change inradiation particle involves multiple steps: approximating anelectron linac with a Bremsstrahlung source, developing a volumephotoneutron source and approximating the emission and energyspectrum of fission signatures.

3. MCNPX simplification methodology

The simulation code, MCNPX, is able to perform complex,full-scale simulations of any radiation inspection systemby following the transport processes of radiation particles.A photoneutron-based active interrogation system detects nucle-ar materials when induced fission signatures are identified by thedetectors. Fig. 1 shows all the key radiation conversion steps thatappear in the interrogation system. The system utilizes anelectron accelerator to generate a Bremsstrahlung beam. TheBremsstrahlung photons then produce photoneutrons in a con-verter, which subsequently induce fission in the nuclear material(NM, such as 235U, 238U, etc.), if present, in the cargo. Finally,

Page 2: Simulation methods for photoneutron-based active interrogation systems

M. King et al. / Nuclear Instruments and Methods in Physics Research A 652 (2011) 129–132130

the fission signatures that traverse the cargo contents aresuccessfully detected. Although MCNPX can readily model andsimulate each of these physical processes, the results would beinadequate unless the particle statistics is reliable and thesimulation is completed in a reasonable amount of time. Forexample, a MCNPX simulation of 1�109 electron particles maytake one hour to generate a statistically meaningful fission rate. Itwould however take 71 years to simulate what a 100 mA electronlinac or 6.25�1014 electrons/s would generate in one second!

For all of the steps shown in Fig. 1, it is highly desirable todevelop reliable simplifications and approximations that willspeed up the overall computation and provide better statisticalaccuracy to the Monte Carlo calculation.

3.1. Bremsstrahlung

An accurate Bremsstrahlung photon spectra approximation isused to bypass the need to simulate electrons, a computationallyexhaustive process in MCNPX. The generated photon sourceshould reflect the fact that the intensity and energy of Brems-strahlung photons are angular dependent. In order to correctlytake this into account, the model tallies the photons emitted froma 9 MeV electron pencil beam incident on a tungsten disc targetwith a copper cooling plate. A spherical tally surface surroundsthe target materials and the photons are tallied in 51 angular binsand 0.1 MeV energy bins. An FRV special treatment card definesthe vector used to define the emitted photon angle as it crossesthe tally surface. Without the FRV vector, which is used in place ofthe vector normal to the surface (a spherical surface in this case),a 01 photon would be incorrectly binned together with a 1801photon. To utilize the approximate Bremsstrahlung spectra as anew source definition, the source card will define the emission as

Electron Linac

Bremsstrahlung

Photoneutrons

Fission Rate

Fission Products

Detector Count Rates

Fig. 1. The cascade chart shows the steps for a photoneutron-based inspection

system.

ConcreteDetector

Lead

DUPhotons

Converter

Detector

Fig. 2. (a) Collimated Bremsstrahlung photons hit the converter creating neutrons that i

without the simulation of photons.

a point source and provide the photon energy spectra andintensity as a function of angle.

3.2. Volume photoneutron source

The method of generating a neutron source, which is devel-oped to further increase fission tally statistics and decreasecomputational time, differs from generating a Bremsstrahlungsource in that the photoneutron source requires the preservationof the initial neutron energy and position, whereas the Brems-strahlung source is emitted as a point source. In order to maintainthe origin and energy of the photoneutron, a volume neutronsource is developed.

3.2.1. Experimental volume neutron source

In laboratory experiments, a block-shaped beryllium piece orheavy water container is placed in front of the linac to producephotoneutrons. Fig. 2 illustrates the difference between simulationsutilizing a photon source vs. a volume neutron source. Fig. 2a showsa collimated photon beam striking the photoneutron converterwhile Fig. 2b shows neutrons being emitted from the convertervolume. Compared to utilizing the photon source, the volumeneutron source can more efficiently determine the fission ratesinside the depleted uranium (DU) slabs.

Preserving the position and energy of the photoneutrons isachieved by a two-step process, where the intensity and energy ofthe neutrons is determined in two separate simulation steps. Thegoal is to emit the neutrons with a constant energy spectrum andwith an intensity that varies along the converter in the directionof the beam. As the photons are attenuated by the converter,fewer and fewer photons are available to create neutrons.To determine the correct energy spectra, neutron energies aretallied for different thicknesses of converter material from 0.01 to1 cm, where the collimated photon beam strikes a large area slice.As the slice thickness increases, the neutron spectra soften inenergy. A correct, unmoderated neutron spectrum is generatedwhen a photon beam strikes an infinitely thin slice of convertermaterial. The energy spectrum emitted from a 0.1 cm thick slice ischosen as a compromise. To tally the dependence of thephotoneutron generation intensity along the thickness of theconverter, i.e. in the direction of the beam, the converter issegmented into the predetermined 0.1 cm thick slices. Becausethe beam is so well collimated, there is no dependence ofthe photoneutron intensity in the perpendicular directions.

ConcreteDetector

Lead

Detector

nduce fission; (b) the volume neutron source emits photoneutrons inducing fission

Page 3: Simulation methods for photoneutron-based active interrogation systems

120°-160°

0°-45°155°175°

36 slices ( 1 cm thick)

Fig. 3. The segmented photoneutron converter emits neutrons with an energy and

intensity that are slice dependent.

Table 1A comparison of simulated neutron fission rates with Be and D2O converters in the

experimental geometry.

Nuclide Be converter D2O converter

Ratio of fission rate

neutron/photon(%)

Ratio of fission rate

neutron/photon(%)

238U 91.2 97.2

DU (0.2% 235U) 94.8 97.1

NU (0.7% 235U) 97.7 99.0

Fig. 4. A 3-dimensional rendering, displayed by MCNPvised [4], of the proposed

scanning system.

Table 2A comparison of simulated neutron fission rate with Be and D2O converters in the

system geometry.

Converter

material

Cargo

contents(g/cc Fe)

Ratio of fission rate

neutron/photon(%)

Be 0.57 82

D2O 0.57 82

M. King et al. / Nuclear Instruments and Methods in Physics Research A 652 (2011) 129–132 131

The volume neutron source is defined as a group of 0.1 cm thickslices with dimensions that correspond to the photon beam shape,an energy spectrum that corresponds to that of a 0.1 cm thickslice and intensity that corresponds to that generated by theappropriate photon intensity along each 0.1 cm step in the beamdirection.

3.2.2. System volume neutron source

For the final interrogation system, a cup-shaped converter isdesigned to surround the linac front-end to maximize the solidangle for photoneutron conversion. The volume neutron sourcesignificantly increases the fission rate statistics in hydrogenouscargos and expedites dose rate simulations where many iterationsof shielding geometries are required. The development ofthe volume neutron source is more complex compared to thedevelopment of the experimental volume source because in theexperiment the collimator only allows a small cone of photons tostrike the converter, which eliminates the photon energy andintensity angular dependence. However, in the final system’sgeometry, the converter subtends almost all photon angles so thephotoneutron intensity and energy spectra vary within theconverter.

The generation of the cup-shape volume neutron sourceinvolves two separate simulations. First, the ideal photoneutronspectrum is determined for every 51 angle, corresponding to the51 angular bins tallied from the Bremsstrahlung spectra. Second,the converter is segmented into vertical and annular segments asshown in Fig. 3. Preferably, the energy of the photoneutronswould also be a function of radii, because each annular segmentgets hit by different Bremsstrahlung angles, but currently, MCNPXdoes not allow two-level dependencies. So, the approximation ofthe neutron energy spectrum for each vertical slice is determinedby averaging the ideal photoneutron energy spectra across all theannular segments of a given vertical slice. Fig. 3 illustratesthe approximation at three vertical slices shown by red lines. TheBremsstrahlung photons that hit the front-most slice are between01 and 451. In this case, the photoneutron spectrum that will beemitted from that front vertical slice is an average of thephotoneutron spectra generated by Bremsstrahlung photonsbetween 01 and 451.

4. Simulation comparison

4.1. Experimental model

To validate the experimental geometry volume neutron sourceby simulation, neutron fission rates are determined in the slabs ofdifferent 235U enrichments by utilizing two different starting sourceparticles, Bremsstrahlung photons and neutrons. The first simulationutilizes Bremsstrahlung photons to generate the photoneutrons,which then induce fission. The second simulation utilizes thevolume neutron source, bypassing the need to simulate photons

and speeding up the simulation, to tally the neutron fission rate.Table 1 demonstrates the excellent agreement between the twomethods. Furthermore, by modeling a narrow range of 235Uenrichments, which emphasizes different parts of the photoneutronenergy spectrum due to the high thermal neutron cross-section, theresults in Table 1 show that the emitted neutron spectrum from thevolume is nearly identical to the much longer simulations startingwith Bremsstrahlung photons.

4.2. Final system model

During a scan, the neutrons emitted by the converter traversethe cargo container inducing detectable fission signatures inhidden nuclear materials. Fig. 4 shows the 3-dimensional modelused to simulate the induced fission rate in 235U from the volumeneutron source.

To validate the proposed system’s volume neutron source,simulation results were compared between using the Brems-strahlung photons as the starting source particle vs. using thevolume neutrons as the starting source particle. This comparisonis equivalent to the comparison that was done for the experi-mental setup. Table 2 shows a good agreement in fission ratesbetween the two methods. Although the agreement is not perfect,the computational time saved greatly outweighs the discrepancy.In order to achieve equivalent neutron fission rate statistics, thephoton simulation takes �1000 times longer to run than thevolume neutron source simulation.

Page 4: Simulation methods for photoneutron-based active interrogation systems

Table 3A comparison of experimental and simulation delayed-g count rate ratios between

the two converters for the scintillation detectors.

(Be/D2O) ratio of delayed-g counts Experimental ratio MCNPX ratio

Scintillator detectors 1.73 1.6

M. King et al. / Nuclear Instruments and Methods in Physics Research A 652 (2011) 129–132132

5. Experimental comparison

To further corroborate the precision of the experimentalvolume neutron source, detector count rates measured in thelaboratory are compared to the simulated detector count rates. Inthe model, fission g-rays are emitted isotropically and withuniform intensity from within the DU slab volume, which are thendetected in the scintillators. The model assumes that the fissionrate intensity is uniform over the entire slab. Shown in Fig. 2 is theexperimental setup that includes two oppositely faced scintillatordetectors, where one is plastic and the other is fluorocarbon. Bothdetectors are used to detect the neutron-induced fission delayedg-rays from the depleted uranium slabs. Table 3 shows a fairlygood agreement between experimental and simulated delayedgcount rate ratios of the two converters.

6. Discussion

A photoneutron-based active interrogation system relies onthe transformation of multiple types of particles in order to detecthidden nuclear materials in cargo containers. Each change inparticle, from electron to photon, photon to neutron and neutron

to fission signatures is so computationally intensive that trying totally detector count rates by starting with electrons in a single-shot simulation would take years to complete. The developmentof the Bremsstrahlung and then the resulting volume neutronsource approximations minimizes computational time andpermits the simulation of denser hydrogenous cargo and difficultnuclear material shielding configurations. The volume neutronsource not only significantly improves the sampling of fissionrates and ultimately the detector count rates but also corrobo-rates experimental data, providing valuable and accurate perfor-mance estimates for the proposed photoneutron-basedinterrogation system.

Acknowledgments

This work is supported by the Department of HomelandSecurity under the Contract HSHQDC-08-C-00125. The firstauthor, Michael King, would like to thank Dr. Tak Pui Lou fromthe Lawrence Berkeley National Laboratory (LBNL) and Dr. JohnHendricks from Los Alamos National Laboratory (LANL) for manyfruitful discussions.

References

[1] T. Gozani, et al., in: AIP Conference Proceedings CAARI2010, 2010, in press.[2] T. Shaw, et al., in: AIP Conference Proceedings CAARI2010, 2010, in press.[3] D.B. Pelowitz, MCNPX User’s Manual, Version 2.5.0, LA-CP-05-0369 ed. (LANL),

2005.[4] L.L. Carter, R.A. Schwarz, J. Pfohl, Trans. Am. Nucl. Soc. 81 (1999) 256.


Recommended