Monte Carlo Particle Transport Calculations Research Articles

Comparing the 1st and 2nd moment approaches to variance reduction in Monte Carlo: Two examples

Variance reduction in Monte Carlo particle transport calculations nearly always relies on the generation of information based on the first moment (viz. the adjoint flux or importance). In this paper we revisit the question of the non-optimal nature of this approach. Whilst it has already been shown to be deficient in duct streaming problems, we look at another kind of problem, namely the activation of nuclides whose cross-sections contain resonances. As a benchmark a second moment-based approach that gives near-optimum results is employed. We find important differences in efficiency between the first and second-moment-based approaches. Some discussion is made on the reasons for these differences.

Development and application of variance reduction technique based on response matrix method in the cosRMC code

In the Monte Carlo particle transport calculations, it is difficult to generate reasonable weight window distributions for the deep penetration problem. This work focused on the generation of importance function by the response matrix method. A built-in importance solver was implemented in the Monte Carlo code cosRMC. In addition, this work improved the iterative process of the weight window of the response matrix method for global variance reduction and realized the accelerated convergence of the global weight window generation. The application of the response matrix method to the local and global variance reduction problems verified the good effect of local variance reduction. The effectiveness of the method has also been further verified in the application of CFETR, where the computational efficiency was 21 times that of the initial analogue simulation, and the time of weight window convergence of the improved response matrix method was reduced by about 24.9%.

Preparation of a water activation experiment at JET to support ITER

Fast neutron induced water activation will result in significant additional nuclear heating loads to cryogenic components in the ITER fusion reactor during operation, which are considered as an important issue in the ITER design. A water activation experiment has been proposed at the Joint European Torus (JET) which would enable measurements of radiation dose rates and 16N concentrations in water, activated in the most representative fusion environment achievable, and extrapolability to ITER conditions. This paper presents a study aimed at establishing the experiment feasibility, in terms of the identification and selection of an experimental location at JET, and the definition of an experimental setup enabling accurate measurements of the quantities of interest. A suitable experimental location in JET was identified in the basement of the JET Torus Hall. The expected 16O(n,p)16N activation rates in the cooling water in the JET tokamak were determined through Monte Carlo particle transport calculations and the induced 16N activity in the water was propagated to the experimental location. Subsequent calculations were performed to determine the expected detector response to 16N γ rays for three different JET pulse scenarios. Two scintillator γ detectors were calibrated with standard γ calibration sources and tested with a 244Cm/13C neutron source, also emitting 6.13 MeV γ rays, corresponding to the 16N γ ray energy. The presented work demonstrates the feasibility of a water activation experiment at JET and provides valuable information for the experiment design.

CEA-JSI Experimental Benchmark for validation of the modeling of neutron and gamma-ray detection instrumentation used in the JSI TRIGA reactor

Constant improvements of the computational power and methods as well as demands of accurate and reliable measurements for reactor operation and safety require a continuous upgrade of the instrumentation. In particular, nuclear sensors used in nuclear fission reactors (research or power reactors) or in nuclear fusion facilities are operated under intense mixed neutron and gamma-ray fields, and need to be calibrated and modeled to provide selective and accurate neutron and gamma-ray measurements. The French Atomic Energy and Alternative Energies Commission (CEA) and the Jožef Stefan Institute (JSI) have started an experimental program dedicated to a detailed experimental benchmark with analysis using Monte Carlo particle transport calculations and a series of neutron and gamma-ray sensor types used in the JSI TRIGA Mark II reactor. CEA has setup a simplified TRIPOLI-4® modeling scheme of the JSI TRIGA reactor based on the information available in the IRPhEP benchmark in order to facilitate analysis of future neutron and gamma-ray measurements. These allow the CEA to perform a TRIPOLI-4 instrumentation calculation scheme benchmarked with the JSI MCNP model. This paper presents the main results of this CEA calculation scheme application and the analysis of their comparison to the JSI results obtained in 2012 with the MCNP5 & ENDF/B-VII.0 calculation scheme. This paper will conclude with some information about the new experimental program to be carried out in 2022 in the TRIGA reactor core.

A Directional Detector Response Function for Anisotropic Detectors

For a radiation detector that is not isotropic, a directional detector response is needed to accurately account for the variation in a detector’s behavior depending on the incoming particle direction. The concept of the detector response function has been extended to include particle direction using a set of pregenerated detector responses based on the orientation of the incoming radiation and the detector. This directional detector response function (DDRF) then can be applied to the flux and current tallies computed by a Monte Carlo simulation. Validation of the new approach has been done by comparing simulated count rates processed with the DDRF to measured count rates taken with a 5.08 × 10.16 × 40.64-cm NaI(Tl) detector. The comparisons show that the applied method produces good agreement with both background and source measurements with a 137Cs source. Furthermore, separation of the detector response generation from Monte Carlo particle transport calculations provides greater flexibility in locating single or multiple detectors without any interference in the model and also enables simulation of various models using the same detector response without the need for generating additional detector responses if the same detector is being used.

A probability-conserving cross-section biasing mechanism for variance reduction in Monte Carlo particle transport calculations

In Monte Carlo particle transport codes, it is often important to adjust reaction cross-sections to reduce the variance of calculations of relatively rare events, in a technique known as non-analog Monte Carlo. We present the theory and sample code for a Geant4 process which allows the cross-section of a G4VDiscreteProcess to be scaled, while adjusting track weights so as to mitigate the effects of altered primary beam depletion induced by the cross-section change. This makes it possible to increase the cross-section of nuclear reactions by factors exceeding 10 4 (in appropriate cases), without distorting the results of energy deposition calculations or coincidence rates. The procedure is also valid for bias factors less than unity, which is useful in problems that involve the computation of particle penetration deep into a target (e.g. atmospheric showers or shielding studies).

SU-FF-T-34: A New Method of Beam Let Weight and Dose Reconstruction for IMRT: Fine Mesh Study

A new method was developed for the reconstruction of beamlet weight and dose in a patient in intensity modulated radiation therapy(IMRT). The method is based on linear relationship between beamlets and dose scoring voxels in EPID, expressed in terms of beam let weights and kernels. Thus, the reconstruction does not involve iteration or assumption of doseimage in EPID as a primary fluence. The kernels quantify dose deposition in each voxel in EPID from each beamlet and are determined by Monte Carlo particle transport calculations. The method was demonstrated computationally on a coronal plane in a phantom. Kernels on this plane and EPID were first calculated for each beamlet. Two IMRTdose distributions of pyramid and inverse pyramid shapes were constructed in phantom and EPID by weighting kernels of each beam let differently. An accident or error was designed by changing the weight of a certain beamlet, modeling MLC uncertainty or output instability. Using the dose difference in EPID due to the accident, the relationship was inversely solved for the changed amount in beamlet weight. For the demonstration, 30 × 12 (6 × 6 cm2) source matrix was used involving the construction of 1296 × 360 kernel matrix in EPID and less than 10 seconds of calculation time and 10 MB of memory were used. The dose change due to the accident in phantom was calculated weighting the predetermined dose kernels in phantom by the reconstructed weight difference. Comparing the reconstructed weight change with the imposed change, the result showed a negligible error due to rounding‐off of data. This validation has shown that the method is practically suitable for the verification of IMRT and useful for application in adaptive radiation therapy. A follow‐up study will be performed, that includes detailed modeling of a therapeutic beam and EPID and experiments.

Ex Post Facto Monte Carlo Variance Reduction

The variance in Monte Carlo particle transport calculations is often dominated by a few particles whose importance increases manyfold on a single transport step. This paper describes a novel variance reduction method that uses a large importance change as a trigger to resample the offending transport step. That is, the method is employed only after (ex post facto) a random walk attempts a transport step that would otherwise introduce a large variance in the calculation.Improvements in two Monte Carlo transport calculations are demonstrated empirically using an ex post facto method. First, the method is shown to reduce the variance in a penetration problem with a cross-section window. Second, the method empirically appears to modify a point detector estimator from an infinite variance estimator to a finite variance estimator.

A new approach to geometry modeling for Monte Carlo particle transport: An application to the EGS code system

A new technique has been developed for Monte Carlo particle transport calculations to handle the geometry for systems having complex structure. In this technique, the geometry of a problem is determined as a superposition of many simple geometries in the same space. The actual geometry of a region is determined by picking between these geometries based upon which one is real and which are virtual at a particular point in space. This method was applied in the EGS code system. The effect of this method on computation time and accuracy of the results are discussed.

A parallel Monte Carlo code for planar and SPECT imaging: implementation, verification and applications in 131I SPECT

This paper reports the implementation of the SIMIND Monte Carlo code on an IBM SP2 distributed memory parallel computer. Basic aspects of running Monte Carlo particle transport calculations on parallel architectures are described. Our parallelization is based on equally partitioning photons among the processors and uses the Message Passing Interface (MPI) library for interprocessor communication and the Scalable Parallel Random Number Generator (SPRNG) to generate uncorrelated random number streams. These parallelization techniques are also applicable to other distributed memory architectures. A linear increase in computing speed with the number of processors is demonstrated for up to 32 processors. This speed-up is especially significant in Single Photon Emission Computed Tomography (SPECT) simulations involving higher energy photon emitters, where explicit modeling of the phantom and collimator is required. For 131I, the accuracy of the parallel code is demonstrated by comparing simulated and experimental SPECT images from a heart/thorax phantom. Clinically realistic SPECT simulations using the voxel-man phantom are carried out to assess scatter and attenuation correction.

Assessment of sensitivities and uncertainties in Monte Carlo particle transport calculations for neutron spectrometry

Monte Carlo particle transport (MCPT) calculations are increasingly used in neutron metrology and especially in neutron spectrometry for the numerical simulation of measurements based on computational models for the underlying experiment. Sensitivities express the dependence of results on input data and uncertainties are a measure of the dispersion of calculated or measured values. The knowledge of sensitivities is instrumental for physical understanding and a possible optimisation of computational models and/or the experiments. The assessment of uncertainties is necessary to state the degree of belief in results and to objectively validate calculations by measurements. Methods used in MCPT calculations which assess sensitivities of the computational model and propagate uncertainties associated with input data are discussed. The technique most often used is demonstrated by a simple example.

A New Weight-Dependent Direct Statistical Approach Model

A weight-dependent capability is inserted into the direct statistical approach (DSA) to optimize splitting and Russian roulette (RR) parameters in Monte Carlo particle transport calculations. In the new model, splitting or RR is carried out on a progenitor arriving at a surface in such a way that the weight of the progeny is fixed (for the particular surface). Thus, the model is named the DSA weight tine model. In the presence of weight-dependent games, all components of the second moment, and the time, are not separable. In the absence of weight-dependent games, the component of the second moment describing the weight-dependent splitting or RR is still not separable. Two approximations are examined to render this component separable under these circumstances. One of these approximations, named the noninteger approximation, looks promising. The new DSA model with the noninteger approximation is tested on four sample problems. Comparisons with the previous weight-independent DSA model and with the MCNP (version 4a) weight window generator are made.

Importance Estimation in Forward Monte Carlo Calculations

AbstractA method for estimating the importance function in forward Monte Carlo particle transport calculations is described. The importance function is estimated for every region of phase-space. Although subject to statistical fluctuations, the estimated importance function has proven to be a very valuable tool for selecting variance reduction parameters. Some deep penetration calculations are included to demonstrate applicability.