Monte Carlo benchmarks

Benchmark Monte Carlo simulations of a different class of athermal polymer mixtures have recently been carried out by Weinhold et al. An equimolar = 0.5), constant-volume binary blend was considered. The polymers were modeled as semiflexible, tangent bead chains of equal degrees of polymerization, N, interacting via a purely hard-core potential... 

The second contribution spans an even larger range of length and times scales. Two benchmark examples illustrate the design approach polymer electrolyte fuel cells and hard disk drive (HDD) systems. In the current HDDs, the read/write head flies about 6.5 nm above the surface via the air bearing design. Multi-scale modeling tools include quantum mechanical (i.e., density functional theory (DFT)), atomistic (i.e., Monte Carlo (MC) and molecular dynamics (MD)), mesoscopic (i.e., dissipative particle dynamics (DPD) and lattice Boltzmann method (LBM)), and macroscopic (i.e., LBM, computational fluid mechanics, and system optimization) levels. 

The need for computer simulations introduces some constraints in the description of solvent-solvent interactions. A simulation performed with due care requires millions of moves in the Monte Carlo method or an equivalent number of time steps of elementary trajectories in Molecular Dynamics, and each move or step requires a new calculation of the solvent-solvent interactions. Considerations of computer time are necessary, because methodological efforts on the calculation of solvation energies are motivated by the need to have reliable information on this property for a very large number of molecules of different sizes, and the application of methods cannot be limited to a few benchmark examples. There are essentially two different strategies. 

We have devoted considerable space to discuss the limitations of the Monte Carlo approach based on an ab initio reaction coordinate, as this approach seems to us to be quite important as a source of benchmarks for other approaches. The limitations we have highlighted can be easily removed, for example, by using ab initio calculations with the continuum approach as a starting point. 

The option-adjusted spread (OAS) is the most important measure of risk for bonds with embedded options. It is the average spread required over the yield curve in order to take into account the embedded option element. This is, therefore, the difference between the yield of a bond with embedded option and a government benchmark bond. The spread incorporates the future views of interest rates and it can be determined with an iterative procedure in which the market price obtained by the pricing model is equal to expected cash flow payments (coupons and principal). Also a Monte Carlo simulation may be implemented in order to generate an interest rate path. Note that the option-adjusted spread is influenced by the parameters implemented into the valuation model as the yield curve, but above all by the volatility level assumed. This is referred to volatility dependent. The higher the volatility, the lower the option-adjusted spread for a callable bond and the higher for a putable bond. 

In this paper, the Subset Simulation (SS) and Line Sampling (LS) methods have been considered for improving the efficiency of Monte Carlo Simulation (MCS) in the estimation of system failure probability. A structural reliability model of hterature, i.e. the cracked plate model, has been taken as benchmark to test the two methods. 

The validation effort was pursued in dose cooperation with CEA Cadarache. Its aims were twofold comparison between our deterministic route (MICROX-2/TWODANT) and the Monte Carlo (MCNP-4A) one, on the one side, and support for the validation effort on the European code system ECCO/ERANOS, on the other side. The work concentrated on three numerical benchmarks derived from the ZONA-2 series of the CIRANO experimental program (performed in the zero power facility MASURCA at Cadarache). For detailed results, see references [5, 6, 7j here only a brief summary of the main findings is given. 

A distinctly different approach, which has witnessed much progress recently, is large scale Monte Carlo and molecular dynamics computer simulations [4]. These studies provide many insights regarding the physics of model polymer fluids, and also valuable benchmarks against which approximate theory can be tested. However, an atomistic, off-lattice treatment of high polymer fluids and alloys remains immensely expensive, if not impossible, from a computational point of view. 

Benchmark calculations of GOBIVA and JEZEBEL have been made in order to compare results with more conventional multigroup Monte Carlo and calculations reported in the literature. As can be seen in Table I, these calculations are in agreement with the consensus of results obtained using other codes with ostensibly the same ENDF/B cross-section sets. 

Table I contains a brief summary of eigenvalue calculations made with VIM-1. They should be regarded as prellmhiary since they were not made with the latest ENDF/B data. They do, however. Indicate that a non-multi oup, fast critical assembly Monte Carlo code can now be used as a tool for analyzing fast critical experiments and that benchmark calculations can now be performed at moderate cost using standardized sets of heterogeneous input data.

Evaluation of Several benchmark reactors has been made with ENDF/B Phase I and Phase n data. These calculations were made using the ESP code. ESP Is a Monte Carlo code designed for general reactor analysis, with an energy range of 10 eV to 15 MeV and capable of handling complicated reactor and reactor cell geometries. Input data to the code must be In the ENDF/B library tape format, and all Phase I and Phase n data applicable.to reactor calculations can be utilized. 

W. ROTHENSTEIN, Monte Carlo Code (REPC) to Calculate Resonance Reaction Rates in Thermal Reactor Lattice Benchmarks, NNCSC Internal Memorandum, Brookhaven National Laboratory (1975). 

Adherence to the Quality Plan and related procedures is expected to reduce the probability of error in calculations by requiring documentation of benchmarks and applications, thereby causing the user to think through the various steps in a calculation, and by requiring independent double checks of all steps of a criticality analysis from reading dimensions on a drawing to interpreting Monte Carlo results. 

Monte Carlo Benchmarking of Fast Reactor Computations, . M. Gelbard, R. E. Prael, B. G. Reynolds (AND, invited... 

The neutron multiplication factor (Keff) of each benchmark core was calculated using the Monte Carlo criticality code KENO IV. The 123-group XSDRN cross-section set was used. Resonance self-shielding was accounted for in the U isotope only. Self-shielding corrections were made with the NIT AWL code from the AMPX package. ... 

Thermal Reactor Benchmark Calculations with the MCNP Monte Carlo Code, Richard E. Prael(LASL)... 

A number of thermal reactor benchmark calculations have been performed using the continuous-energy Monte Carlo code MCNP with ENDF/B-IV data. The principal motivations for these calculations were ... 

J. HARDY, Jr., Analysis of Homogeneous Plutonium Benchmark Assemblies with the RCPOl Monte Carlo Program and ENDF/B-IV Data, Private Communication to CSWEG Thermal Data Testing Subcommittee (Apr. 1979). 

The mafiionatical-criticality analyses were performed using the KENO Monte Carlo computer program together with the 123-groiq> GAM THERMOS neutron cro -s6ction set. Fuel pin cdl k, Calculations were performed to establidi the most reactive lattice pitch as a frmetion of U enrichment and bonm concentrations. Three-dimensional (3-D) KENO keff calculations were used to establish the reactivity of tile reactor under a variety of states. Since the TMI-2 coolant will eventually reach room temperature, all criticality analyses (with the exception of the hot, clean benchmark TMI-2 conjuration) were performed at this most reactive (neutronically) temperature. The moderator density was taken as 1.0 g/cm and the fuel (UOj) density was assumed as 95% theoretical. 

---