A. Haghighat

Automated variance reduction of Monte Carlo shielding calculations using the discrete ordinates adjoint function

Although the Monte Carlo method is considered to be the most accurate method available for solving radiation transport problems, its applicability is limited by its computational expense. Thus, bia...

Effectiveness of pentran™'s unique numerics for simulation of the Kobayashi benchmarks

In this paper, we use the PENTRAN (Parallel Environment Neutral-particle TRANsport), 3-D parallel Sn code to solve three 3-D benchmark problems. Each of the problems contains three regions: source, void, and shield. Two shield materials, zero (pure absorber) and 50% scattering ratios, are tested. This combination of a discontinuous source, void, and low scattering shield is highly problematic for the Sn method that uses a limited number of directions. In this paper, we demonstrate that the unique numerical formulations and features of PENTRAN provide the capability of obtaining relatively accurate solutions for such situations. For the pure absorber cases, we obtain maximum differences from the analytical solutions of < 40%, 24%, and 20% for the problems 1 to 3, respectively. For the 50% scattering cases, these differences reduce to <20% for all problems. In most cases, the maximum error occurs at large distances from the source. We have demonstrated that in order to obtain these solutions, it is necessary to use three important features of PENTRAN, including variable meshing, Taylor Projection Mesh Coupling (TPMC), and the adaptive differencing strategy with the DTW and EDW differencing schemes.

Effects of SN method numerics on pressure vessel neutron fluence calculations

An accurate prediction of the reactor pressure vessel (PV) fast neutron fluence (E > 1.0 MeV or E > 0.1 MeV) is necessary to ensure PV integrity over the design lifetime. The discrete ordinates method (S N method) is the method of choice to treat such problems, and the DORT S N code is widely used as a standard tool for PV fluence calculations. The S N numerics and the corresponding DORT numerical options and features offer alternative choices that increase flexibility but also impact results. The effects of S N numerics based on PVfluence calculations for two pressurized water reactors are examined. The differencing schemes [linear, zero-weighted (ZW), and θ-weighted (TW)] and their interactions with spatial and angular discretization are also examined. The linear and TW (θ = 0.9) schemes introduce unphysical flux oscillations that for certain groups and positions may exceed 10%. The ZW scheme produces smooth results ; however, its results differ from the other two schemes. A good compromise for PV fluence calculations is a TW scheme with a small θ value (i.e., θ = 0.3), which reduces the uncertainty to ∼3%. Angular discretization and spatial mesh size employed in typical calculations introduce another ∼3 and ∼2% uncertainty, respectively. The analysis further shows that the fixup is not necessary for the negative scattering source. The pointwise convergence criterion is also not a critical issue in the fast energy range because of a relatively fast convergence rate. Similarly, acceleration parameters impact mainly the execution time and only marginally the results. The root-mean-square combined uncertainty for standard PV fluence calculations due to the options analyzed is ∼5%.

Performance of PENTRANTM 3-D Parallel Particle Transport Code on the IBM SP2 and PCTRAN Cluster

This paper discusses the algorithm and performance of a 3-D parallel particle transport code system, PENTRAN™ (Parallel Environment Neutral-particle TRANsport). This code has been developed in F90 using the MPI library. Performance of the code is measured on an IBM SP2 and a PC cluster. Detailed analysis is performed for a sample problem, and the code is used for determination of radiation field in a real-life BWR (Boiling Water Reactor). Using 48 IBM-SP2 processors with 256 Mbytes memory each, we have solved this large problem in ~12 hours, obtaining a detailed energy-dependent flux distribution.

A New Investigation of Iron Cross Sections via Spherical-Shell Transmission Measurements and Particle Transport Calculations

We are engaged in a multi-year project to study neutron scattering interactions in iron, the principal objective of which is to investigate the well-known deficiency that exists in reactor pressure vessel neutron fluence determinations. Specifically, we are using the spherical-shell transmission method, employing iron shells with different thicknesses, and neutron time-of-flight measurements of the scattered neutrons, in an effort to precisely determine specific energy regions over which deficiencies in the non-elastic scattering cross section for neutron scattering in iron appear to exist. The analysis of the experimental data involves correlating the data with theoretical calculations of neutron transport through the iron spheres in order to evaluate the degree to which the calculated neutron spectra predict the measured spectra relative to different types of particle interactions. In doing so, we have developed new methodologies for performing neutron transport calculations that will be useful to a range of transport problems. Preliminary results show good agreement between the experimental data and the calculated distribution of neutron flight times over much of the data range, except for the contribution due to breakup neutrons.

New Particle Transport Methods for Design and Optimization of Spherical-Shell Transmission Measurements

This paper discusses new particle transport methods developed for accurate investigation of iron non-elastic scattering cross sections using the spherical-shell transmission method, employing iron shells with different thicknesses, and neutron time-of-flight spectroscopy of the scattered neutrons. New calculational techniques based on the deterministic and Monte Carlo methods have been developed for design and optimization of the experiment, and for identification and reduction of experimental uncertainties. The new methods include a new tallying option for the MCNP code and new quadrature sets for the 3-D parallel Sn code PENTRAN. The new tallying option is used to determine the optimum source energy versus target thickness, and among the new quadrature sets, the Pn-Tn with ordinate-splitting technique has resulted in accurate flux distributions as compared to Monte Carlo prediction.