G. Sjoden

Electron Dose Kernels to Account for Secondary Particle Transport in Deterministic Simulations

Abstract For low-energy photons, charged-particle equilibrium usually exists within the patient treatment volume, in which case the photon absorbed dose D is equal to the collisional kerma Kc; however, this is not true for the dose buildup region near the surface of the patient or at interfaces of dissimilar materials, such as tissue/lung, where corrections for secondary electron transport may be significant. This is readily treated in Monte Carlo codes, yet difficult to treat explicitly in deterministic codes due to the large optical thicknesses and added numerical complexities in reaching convergence in photon-electron transport problems. To properly treat three-dimensional electron transport physics deterministically, yet still achieve reasonably fast and accurate whole-body computation times using high-energy photons, angular-energy-dependent transport “electron dose kernels” (EDK-SN) have been developed. These kernels were derived via full physics Monte Carlo electron transport simulations and are applied using scaling based on rapid deterministic photon solutions over the problem phase-space, thereby accounting for the dose from charged-particle electron transport. As a result, accurate whole-body doses may be rapidly achieved for high-energy photon sources by performing a single deterministic SN multigroup photon calculation on a parallel cluster with PENTRAN, then linking the SN-derived photon fluxes and net currents to Monte Carlo–based EDKs to account for a full physics dose. Water phantom results using a uniform 0- to 8-MeV step uniform beam indicate that the dose can be accurately obtained within the uncertainty of a full physics Monte Carlo simulation. Followup work will implement this method on phantoms.

Benchmarking of PENTRAN-SSN Parallel Transport Code and FAST Preconditioning Algorithm Using the VENUS-2 MOX-Fueled Benchmark Problem

The discrete ordinates method (Sn) is the most widely used technique to obtain numerical solutions of the linear Boltzmann equation, and therefore to evaluate radiation fields and dose rates in nuclear devices. However, it is well known that this method suffers from slow convergence for problems characterized by optically thick media and scattering ratio close to unity. To address this issue we have developed a new preconditioning algorithm based on the even-parity simplified Sn (EP-SSN) equations. The new method is based on the flux acceleration simplified transport (FAST) algorithm which is implemented into the PENTRAN-SSN code system. The code system is designed for parallel computing architectures; PENTRAN-SSN features spatial, angular, and energy domain decomposition algorithms. The FAST preconditioner is parallelized with a spatial domain decomposition algorithm. In this paper, our objective is to test the performance of the new preconditioning system for a three-dimensional shielding calculation based on the VENUS-2 MOX-fueled benchmark problem, issued by OECD/NEA (Organization for Economic Co-operation and Development/Nuclear Energy Agency.

Positive SNM Gamma Detection Achieved Through Synthetic Enhancement of Sodium Iodide Detector Spectra

We have developed a new algorithm, ASEDRA, to post-process scintillator detector spectra to render photopeaks with high accuracy. ASEDRA, or ldquoAdvanced Synthetically Enhanced Detector Resolution Algorithm,rdquo is currently applied to NaI(Tl) detectors, which are robust, but suffer from poor energy resolution. ASEDRA rapidly post-processes a NaI(Tl) detector spectrum over a few seconds on a standard laptop without prior knowledge of sources or spectrum features. ASEDRA incorporates a novel denoising algorithm based on an adaptive Chi-square methodology called ACHIP, or ldquoAdaptive Chi-quare Processed denoising.rdquo Application of ACHIP is necessary to remove stochastic noise, yet preserve fine detail, and can be used as an independent tool for general noise reduction. Following noise removal, ASEDRA sequentially employs an adaptive detector response algorithm using detector Monte Carlo data to remove the spectrum attributed to specific gammas. Tests conducted using a 2rdquo times 2rdquo NaI(Tl) detector, along with a HPGe detector demonstrate the accuracy of ASEDRA for both photopeak identification and relative yield. In this paper, we present successful results for both WGPu and natural uranium Special Nuclear Materials (SNM) sources, comparing key photopeak energies and intensities to known values. Moreover, the denoising and synthetic resolution enhancement algorithms can be adapted to any detector. ACHIP and ASEDRA are covered under a Provisional Patent, Registration Number #60/971,770, 9/12/2007.

CRITICAL DISCRETIZATION ISSUES IN 3-D SN SIMULATIONS RELEVANT TO DOSIMETRY AND MEDICAL PHYSICS

Abstract Dosimetry problems inherently involve dose determinations among widely varying materials and densities, and may require complex, detailed investigations of the angular, spatial, and energy behavior of the applied radiation transporting throughout the simulation geometry. Traditionally, Monte Carlo codes have been implemented in solving these types of problems using voxelized geometries and phantoms. The motivation of this work is to investigate the discretization requirements for deterministic radiation transport simulations for these problems via direct solutions of the linear Boltzmann transport equation, focusing on the discrete ordinates (SN) method. The SN method can yield accurate global solutions, provided the inherent discretizations among the angular, spatial, and energy domains properly represent problem physics. In this paper, the SN approach is implemented using a three-dimensional (3-D) 60Co photon transport simulation to highlight the critical issues encountered in performing deterministic photon simulations in dosimetry problems. Calculations were performed using the PENTRAN parallel SN code to obtain a 3-D distribution of flux and dose computed using a collisional kerma approximation. For an acceptable result, we determined that a minimum angular Legendre-Chebychev quadrature of S32 with P3 anisotropy is required, with block-adaptive meshes on the order of 1 cm, even in air regions, implemented with an adaptive differencing scheme (implemented in the PENTRAN code) to yield optimal solution convergence. Also, photon cross-section libraries should be carefully evaluated for the problem studied; for our test problem, the BUGLE-96 photon library yielded the closest results to Monte Carlo (MCNP5) among those tested. Overall, this work details the levels of discretization involved in performing deterministic computations in dosimetry problems and will be useful in enabling future efforts to perform rapid deterministic computations of phantom doses.

On Neutron Spectroscopy Using Gas Proportional Detectors Optimized by Transport Theory

Abstract We explore in this study the practical limits in designing a neutron detector array to resolve the spectra from special nuclear material (SNM) neutron sources using 3He detectors. We demonstrate that radiation transport analysis yielded a spectrum unfolding strategy based on the energy structure of the BUGLE-96 cross-section library, with 47 neutron energy groups. The initial computational model used is an isotropic planar source incident on a block of high-density polyethylene moderator. Assuming 3He is diluted throughout the moderator, the 3He(n,p) reaction rate energy group matrix in the block was computed using a completely “flat” neutron source spectrum. Analyzing the energy group matrix, there are neutrons from specific collections of energy groups (energy “bands”) that induce a maximum reaction rate in specific locations; we determined that these groups cannot be further differentiated within the energy band using 3He detectors. It was determined that optimal spectral fidelity for SNM detection and characterization is achievable using four spectral bands spanning among groups 1 through 29 (31.8 keV to 17.3 MeV). Using ideal-filter materials to remove the neutrons from different regions of the spectrum, we predicted the maximum neutron spectral resolution obtainable using this approach. To demonstrate our method, we present the optimally detected spectral differences between SNM materials (plutonium and uranium), metal and oxide, using ideal-filter materials. We have also selected a number of candidate filtering materials and, by replacing the ideal filters with real materials, we exemplified our approach with a design of a neutron detector array capable of resolving the spectra from SNM neutron sources using 3He detectors.

Improved plutonium identification and characterization results with NaI(Tl) detector using ASEDRA

The ASEDRA algorithm (Advanced Synthetically Enhanced Detector Resolution Algorithm) is a tool developed at the University of Florida to synthetically enhance the resolved photopeaks derived from a characteristically poor resolution spectra collected at room temperature from scintillator crystal-photomultiplier detector, such as a NaI(Tl) system. This work reports on analysis of a side-by-side test comparing the identification capabilities of ASEDRA applied to a NaI(Tl) detector with HPGe results for a Plutonium Beryllium (PuBe) source containing approximately 47 year old weapons-grade plutonium (WGPu), a test case of real-world interest with a complex spectra including plutonium isotopes and 241Am decay products. The analysis included a comparison of photopeaks identified and photopeak energies between the ASEDRA and HPGe detector systems, and the known energies of the plutonium isotopes. ASEDRAs performance in peak area accuracy, also important in isotope identification as well as plutonium quality and age determination, was evaluated for key energy lines by comparing the observed relative ratios of peak areas, adjusted for efficiency and attenuation due to source shielding, to the predicted ratios from known energy line branching and source isotopics. The results show that ASEDRA has identified over 20 lines also found by the HPGe and directly correlated to WGPu energies.

3D computational and experimental radiation transport assessments of Pu-Be sources and graded moderators for parcel screening

The Florida Institute for Nuclear Detection and Security (FINDS) is currently working on the design and evaluation of a prototype neutron detector array that may be used for parcel screening systems and homeland security applications. In order to maximize neutron detector response over a wide spectrum of energies, moderator materials of different compositions and amounts are required, and can be optimized through 3-D discrete ordinates and Monte Carlo model simulations verified through measurement. Pu-Be sources can be used as didactic source materials to augment the design, optimization, and construction of detector arrays with proper characterization via transport analysis. To perform the assessments of the Pu-Be Source Capsule, 3-D radiation transport computations are used, including Monte Carlo (MCNP5) and deterministic (PENTRAN) methodologies. In establishing source geometry, we based our model on available source schematic data. Because both the MCNP5 and PENTRAN codes begin with source neutrons, exothermic (α,n) reactions are modeled using the SCALE5 code from ORNL to define the energy spectrum and the decay of the source. We combined our computational results with experimental data to fully validate our computational schemes, tools and models. Results from our computational models will then be used with experiment to generate a mosaic of the radiation spectrum. Finally, we discuss follow-up studies that highlight response optimization efforts in designing, building, and testing an array of detectors with varying moderators/thicknesses tagged to specific responses predicted using 3-D radiation transport models to augment special nuclear materials detection.

A New Synthetic Acceleration Technique Based on the Simplified Even-Parity SN Equations

Abstract The source iteration method or Richardson method is the most common technique utilized to solve the linear Boltzmann equation. However, this iterative method exhibits slow convergence for problems with optically thick regions and scattering ratio close to unity. Hence, an effective acceleration technique is necessary for achieving a converged solution in a reasonable amount of time. In this paper, we develop a new synthetic acceleration technique based on the simplified even‐parity SN (SEP‐SN) equations for the discrete ordinates method (SN). We have measured the effectiveness of the SEP‐SN acceleration technique for both fixed‐source and eigenvalue (k‐effective) problems.

SU‐GG‐T‐408: Validation of a Novel Dose Calculation Approach for Heterogeneous Voxelized Phantoms in a Parallel Computation Environment Using Electron Dose Kernels for Radiotherapy

Purpose Here we present a rapid whole bodydose estimation approach applied to a clinical water phantom and whole body CT‐voxelized phantom, using electron dose kernels coupled with deterministic photon transport. This method yields fast, accurate whole body doses.Method and Materials A novel dose calculation methodology called EDK‐SN, or “Electron Dose Kernel‐Discrete Ordinates” rapidly estimates organdoses in a voxelized human phantom, accounts for in‐ and out‐of‐field doses using external photon beam therapy. We begin by solving the complete photon transport problem using parallel computing with the 3‐D discrete ordinates (SN) photon transport code PENTRAN. We then project pre‐computed (via Monte Carlo) voxel‐based Electron Dose Kernels (EDKs), mapping them to surrounding voxels via quaternion rotation, scaled by the magnitude of photon fluence from the SN calculation. An 8 MV flat‐weighted beam is incident on an 11×11×11 cm3 water phantom, and on a 15 year old human phantom, down‐sampled to 1×1×1 cm3 (60×27×167 voxels); a 6 MV Philips Elekta Linacphoton spectrum has also been simulated. The percent depth dose was compared to clinical CC04 chamber measurement results; comparison of doses using the EDK‐Sn method and clinical treatment planning system (in field dose) will also be presented. Results The EDK‐SN technique has demonstrated independent agreement with Monte Carlo photon‐electron transport calculations for whole body dose. The EDK‐SN method yields a speedup of ∼8 (30 minutes versus 4+ hours) over the traditional parallel Monte Carlo, with <7% difference in different organs (smaller given stochastic uncertainties). The Monte Carlo simulated percent depth dose and clinical chamber PDD measurement agree within 10% among different field sizes.Conclusion The EDK‐SN method for high energy photon external beam dose calculations has been validated based on clinical external therapy beam calculations. This method will help to determine both in‐field and out‐of‐field radiationdose for radiotherapy.

MO-E-332-01: DXS a Diagnostic X-Ray Spectra Generator

Purpose: To numerically generate radiographic x‐ray spectra that can be conveniently employed in radiation transport simulations or other radiation detection applications. Method and Materials: Based initially on the Tucker, et al model, we developed and evaluated a new code, DXS (Diagnostic X‐Ray Spectra), to numerically generate spectra for tungsten‐target x‐ray tubes spanning the radiographic energy range. The model parameters in our code were adjusted by comparison with corresponding MCNP5 simulated spectra; we modified the semi‐analytical formulation for the characteristic x‐ray production, a caveat of Tuckers model, by incorporating a factor that better accounts for the dependence of the K‐peaks on the tube potential. Parametric fitting functions are used to model the self‐attenuation in the target and attenuation due to inherent and added filtration (aluminum,beryllium,copper,tantalum are the options implemented in the code), as well as for the tungsten mass stopping power and the Thomson‐Whiddington constant. Comparison with Monte Carlo simulated and published measured spectra were used to validate the new code. Results: Normalized to unit area DXS code‐generated spectra for several tube potentials from 50 to 140 kVp agree well, less than 2% relative difference in nearly all energy bins (2 keV), with corresponding MCNP5 simulated spectra for similar tube parameters. Few exceptions are noted and may be attributed to either poorer statistics in the low and high energy tails of the spectrum, or to insufficient accuracy of the numerical computations for the steepest part of the spectra at high accelerating potentials. Good agreement is seen between the DXS and Bhat et al measured spectra. Conclusion: The DXS code generates the spectra, according to user specified input parameters (tube potential, anode angle, filtratation) and energy intervals, and augments them into any discretized energy group structure. Hence, the code can be of great benefit in radiation transport simulations.