Computational and Experimental Benchmarking for Transient Fuel Testing:   Task 1

T. Downar W. Martin  University of Michigan

C. LeeArgonne National Laboratory

K. SunMIT

May 24, 2016

Task 1• Objective: A comprehensive evaluation of existing TREAT Facility neutronics data using the next generation reactor core neutronics codes.   This will be performed in accordance with  established guidelines per the International Handbook of Evaluated Reactor Physics Benchmark Experiments (IRPhEP). 

• Neutronics Codes:• Monte Carlo:  

• SERPENT (UM)• MCNP  (ANL)• OPENMC (MIT)

• Deterministic:   • PROTEUS   DOE NEAMS  (ANL)• PARCS  US NRC   (UM)

• Benchmarks (UM)• Steady‐State – Two steady state condition benchmarking tests will be selected and studied.• Transient – Two transient condition benchmarking problems will be selected and studied.

Task 1.1 (Steady‐State)  ScheduleTask # Task Title Sub‐Task Owner

1. Neutronics Benchmark Task Lead – T. Downar, UM

1.1 Steady State (SS)

1.1.1 Survey candidate problems T. Downar, UM

1.1.2 Preliminary SS modeling of candidate problems T. Downar, UM

1.1.3 Down‐select to two  problems for benchmark evaluation T. Downar, UM

1.1.4 SS modeling with deterministic U.S. NRC codes PARCS/AGREE T. Downar, UM

1.1.5 SS modeling with deterministic NEAMS code PROTEUS C. Lee, ANL

1.1.6 SS modeling with Monte Carlo code OPENMC  K. Sun, MIT

1.1.7 Comparison of experimental data & model results  T. Downar, UM

1.1.8 Benchmark level evaluation of selected problems T. Downar, UM

1.1.9 Evaluation of uncertainties in selected problems T. Downar, UM

1.1.10Preparation of IRPhEP documentation

T. Downar, UM

1.1.11Submission of SS benchmark for peer review

T. Downar, UM

Task 1.1.3    Downselect to Two Problems for Benchmark Evaluation 

Minimum Critical Core M8CAL

TREAT Benchmark Specifications:    Geometry and Compositions  

UM Serpent Core Model

Model from Batman Report1 SERPENT Model – Top / Axial

Infinite Lattice UQBatman Report

Boron impurity study

Iskenderian, H. P. (n.d.). Post criticality studies on the TREAT reactor (p. 7, Tech.). (NTIS No. ANL‐6115) 50 samples total of 1.25g out of 2.6tons of fuel

• Direct average: • μ 7.8 , σ 1.4• Inverse variance weighting:• μ . , .

Original Calcs on MCC with the SERPENT Core Model  

Case K‐eff Diff (pcm)

Base: 5.9ppm Boron59%GraphitizationENDF/B‐7.1267ppm Fe16 Zr Assembly

1.01846 ±23pcm ‐

7.6ppm Boron600ppm Fe

1.00130±19pcm 1716

100%Graphitization1.00394 ±23pcm 1452

0 Zr Assembly1.01639±21pcm 207

9

Boron Contamination:  Chi‐Square weighting

Group No. Sample size Degree of Freedom Probability of getting 0.9≤ ≤1.1

1 8 7 0.145

2 10 9 0.165

3 20 19 0.241

4 12 11 0.183

• wi=∑

• Weighted mean: 7.53ppm• Weighted std: 1.16ppm

Parametrics with the SERPENT MC Core Model  

Case keff Diff (pcm)

Base: 7.53ppm Boron267ppm Fe59%GraphitizationENDF/B‐7.116 Zr Assembly

1.00413±20pcm ‐

5.9ppm Boron 1.01846±23pcm 1433

600ppm Fe 1.00195 ±19pcm ‐218

0% Graphitization100% Graphitization

1.01356±20pcm0.98950±20pcm

9431463

0 Zr Assembly 1.00214±21pcm ‐199

11

M8CAL SERPENT Results

12

Case keffBase: 7.53ppm Boron

267ppm Fe59%GraphitizationENDF/B‐7.1

C/S Rod 22inCR Rod  OUTTR Rod OUT

1.00394±20pcm

0

2

4

6

8

10

0 10 20 30 40 50 60

Reactiv

ity (%

)

Rod Position (in)

EXP

SERPENT

IRPhEP Benchmark Experiment DocumentationSection 1 Description of the Experiment

A detailed description of the experiments and all relevant experimental data will be provided in the appropriate subsections of section 1. 

Section 2 Evaluation of Experimental DataMissing data or weaknesses and inconsistencies in published data will be discussed and resolved in the appropriate subsections of section 2. The effects of uncertainties in parameter data on the measurement results will be discussed and quantified. Codes and modelling methods used for calculations of the effects will be specified and the use of data with large uncertainties or data that require assumptions on the part of the evaluator will be justified.

Section 3 Benchmark SpecificationsBenchmark specifications will be provided which will include all the data necessary to construct calculationalmodels that best represent the experiment. 

Section 4 Calculated ResultsCalculated results obtained with the benchmark‐model specification data given in Section 3 will be tabulated in this section. These will be regarded as sample calculation and methodologies used for the sample calculations and any other recommendations for the calculations will be described.

Section 5 References / AppendicesAppendix A will provide a description of the options, cross section data, and an input listing for the codes used in the calculations of the results given in Section 4.

Task 1.1.5    Accomplishments at ANL • Preliminary calculations of TREAT benchmark problems (MinCC and M8CAL) using PROTEUS

• Mesh generation for 2D and 3D MinCC and M8CAL cores using CUBIT + UFmesh (built‐in mesh generation of PROTEUS)

• PROTEUS (MOCEX solver) calculations using 9‐group cross sections (generated from Serpent/GenISOTXS) or the cross section API of PROTEUS (ongoing)

• Discussion with the UM team on• IRPhEP benchmarks and documentation• Serpent (UM) and MCNP (ANL) results on MinCC

14

Minimum Critical Core

M8CAL

mesh

Task 1.16 MIT: OpenMC Infinite Fuel Lattice

(1) Outgas Hole

(1)(2)

(3)

(4)

(5)

(2) Upper Fission Gas Vent

(5) Graphite Reflector

(3) Fuel Region

(4) Lower Fission Gas Vent

MIT: OpenMC Mini. Critical Core (MCC)

133 Standard Fuel Assembly 8 Control Rod Fuel Assembly (Fully Withdrawn) 16 Zircaloy Clad Dummy Assembly

MIT: Results for Benchmark CasesNo. Lattice Case k_inf Reactivity (pcm)

0.

Benchmark: ENDF/B‐7.17.53 ppm Boron and 267 ppm IronNo Hydrogen in Fuel< 100 ppm Hafnium in Zr‐3 Clad100% Graphitization 

1.42450 ± 0.00020 N/A

1. Benchmark + 0% Graphitization(No Sab for the graphite in “Fuel”) 1.43852 ± 0.00023 + 684.2

No. MCC Case k_eff Reactivity (pcm)

0.

Benchmark: ENDF/B‐7.17.53 ppm Boron and 267 ppm IronNo Hydrogen in Fuel< 100 ppm Hafnium in Zr‐3 Clad100% Graphitization 

1.00266 ± 0.00016 N/A

1. Benchmark + 0% Graphitization(No Sab for the graphite in “Fuel”) 1.02690 ± 0.00009 + 2354.2

Next Step: M8CAL Core Modeling using OpenMC

Task 1.2 (Transient)  Schedule1.2

Transient (TR)

1.2.1 Survey available TREAT TR data for benchmark problem T. Downar, UM

1.2.2 Preliminary TR modeling of candidate problems T. Downar, UM

1.2.3 Down‐select to two  problems for benchmark evaluation T. Downar, UM

1.2.4 Perform TR modeling with deterministic U.S. NRC codes PARCS/AGREE T. Downar, UM

1.2.5 Perform TR modeling with deterministic NEAMS code PROTEUS C. Lee, ANL

1.2.6 Perform TR modeling with Monte Carlo code OPENMC W. Martin, UM

1.2.7 Benchmark level evaluation of selected problems T. Downar, UM

1.2.7 Evaluation of uncertainties in selected problems T. Downar, UM

1.2.8 Preparation of IRPhE Documentation T. Downar, UM

1.2.9 Submission of TR benchmark for peer review T. Downar, UM

TASK 1.2.6   Transient Modeling with Monte Carlo Code OPENMC      

19

Task Participants Bill Martin (lead) Dr. Scott Wilderman (Research staff) Ethan Pacheck (PhD student) Assistance from Volkan Seker Ben Betzler (ORNL) has agreed to provide assistance to this project.

Have adapted OpenMC to use OTF capability currently in MCNP6. This will account for Doppler broadening during the transient, which will be especially important for LEU.

Using dummy isotope approach suggested by Volkan, have shown that we can model a mixture of graphitized and non-graphitized carbon with OpenMC.

This dummy isotope approach allows OpenMC to account for thermal spectrum feedback due to temperature change when S(a,b) is used. This works because the S(a,b) cross sections are generated at a specific temperature so dummy isotopes can be set up at arbitrary temperatures and then mixed.

TASK 1.2.6   Transient Modeling with Monte Carlo Code (cont.)  

20

Completed specification of a simple benchmark configuration to allow comparison of alternative transient methodology with true time-dependent Monte Carlo. This benchmark configuration has the following features: 3x3 homogenized assemblies Infinite height No air gaps or other features Will allow time-dependent Monte Carlo simulation with OpenMC for

comparison to the alternative methodology developed by Ben Betzler for his PhD thesis.

Preliminary results of test cases are in reasonable agreement with MIT and UM results for the heterogeneous configuration.

Questions?