pebble bed reactor modeling using...

Pebble bed reactor modelingusing Serpent

Heikki Suikkanen, Ville Rintala

Lappeenranta University of TechnologyP.O. BOX 20, FI-53851, Lappeenranta, Finlandphone: +358-503059242, [email protected]

Serpent International Users Group Meeting,Dresden, Germany, September 15, 2011

ContentIntroduction

Pebble bed reactor

Available models in SerpentParticle fuel models in SerpentPebbles in the coreAdvantages of Serpent in pebble bed calculations

ASTRA criticality calculationsASTRA critical experimentsSerpent model of ASTRACriticality calculation results

Thermal-hydraulics couplingReactor physics modelThermal-hydraulics modelCoupling codeRemaining obstacles and development areasTest case

ConclusionsSerpent International Users Group Meeting, Dresden, Germany, September 15, 2011 3 / 25

Introduction

Serpent International Users Group Meeting, Dresden, Germany, September 15, 2011 4 / 25

Introductiony Lappeenranta University of Technology is focusing on

the modeling of pebble bed reactors as a part ofNETNUC (New Type Nuclear Reactors) project fundedby the Academy of Finland.

y Modeling tools are developed and tested to model thereactor physics, thermal-hydraulics and behavior of fuelpebbles.

y Serpent is currently used in the reactor physicsmodeling.


Pebble bed reactor (PBR)

Figure: Schematics of a coated fuelparticle and a cut-in-half fuel pebble.

Figure: Pebbles inside an annular corecavity.


Available models in Serpent


Particle fuel models in Serpenty There are basically three ways

to model particle fuel inSerpent.

1. A regular lattice of particlesinside a pebble.

2. An implicit approach whereparticles are sampled duringthe random walk.

3. All particles are givenposition coordinates in aseparate input file.

Figure: Coated fuel particles explicitlydefined inside a pebble.


Pebbles in the corey The pebbles could be put in a

regular lattice, which is the usualapproach in MC simulations.

y In Serpent, the positions ofpebbles can be given in aseparate input file the same wayas coated particles.

y Defining the pebbles in a denserandom configuration requiressome effort.

y Discrete Element Method (DEM)can be used to create a veryrealistic pebble bed.

Figure: Stochastic configuration ofpebbles inside an annular cavity.


Advantages of Serpent in pebblebed calculations

y HTGR specific geometry types and methods allow easymodeling.

y Delta-tracking provides speedup in geometries with lotsof details.

y Packed beds created with DEM can be used directly.


ASTRA criticality calculations


ASTRA critical experimentsy Five ASTRA critical pebble bed

experiments that were done atthe Kurchatov Institute during2003-2004 and documented inthe IRPhEP Handbook arecalculated.

y A cylindrical steel vessel wheregraphite blocks wereassembled to form anoctagonal annular cavity for thefuel spheres.

Figure: ASTRA annular core setup.


Building the modely A very detailed geometry model is

built based on the available data inthe benchmark documentation.

y DEM is used to pour the pebblesinside the cavity.

y Serpent reads in the pebble positioncoordinates.

y Coated fuel particles are modeled indetail and explicitly defined inside thepebbles.

Figure: Pebbles inside the ASTRA cavity.


Serpent Geometry

Figure: Horizontal cut of the ASTRAgeometry model.

Figure: Vertical cut of the ASTRAgeometry model.


Criticality calculation resultsy Calculations result in

approximately 1000 pcmoverestimation.

y Performed sensitivity analysesdon’t reveal the cause.

y Similar overestimation has beenwitnessed by others e.g. withMCNP.

y Testing of different cross-sectionlibraries is anticipated in thefuture, especially JENDL-4.

y One case took 16 hours with fiveIntel Core i7 processors and 36parallel tasks in total.

Table: Multiplication factors calculated bySerpent.

keff± 1σ keff± 1σCore (experiment) (simulation)1 1.0000 ± 0.0036 1.01052 ± 0.000082 1.0000 ± 0.0036 1.01040 ± 0.000083 1.0000 ± 0.0036 1.01086 ± 0.000084 1.0000 ± 0.0036 1.01005 ± 0.000085 1.0000 ± 0.0036 1.01096 ± 0.00008

Table: Results of the sensitivity analysisof core one.

Changed parameter keff± 1σCore height 182.30 cm 1.00950 ± 0.00008

JEFF-3.1 1.01084 ± 0.00008ENDF/B-VII 1.01112 ± 0.00008Graphite impurities 1.1 1.00777 ± 0.00008ppm boron


Thermal-hydraulics coupling


Thermal-hydraulics (T-H) couplingy Aim is in steady state full-core PBR calculations.

y Due to helium coolant, there are no phase changes...

y ...but the exceptional geometry sets new challenges.

y For T-H, the general-purpose CFD-code ANSYS Fluentis used.

y A coupling code connecting Serpent and Fluent iswritten in Perl.


Reactor physics modely Geometry is modeled like in the

ASTRA case.

y Temperatures are changed bythe internal Dopplerbroadening routine of Serpent.

y Serpent writes the pebble-wisepower data as the result.

−100

−50

0

50

100 0

50

100

150

200

0

50

100

150

200

250

300

350

400

0

0.2

0.4

0.6

0.8

1

Rel

ativ

e po

wer

Figure: Half of a pebble bed showing thepower distribution.


Thermal-hydraulics modely Solving the flow and heat

transfer in detail with CFD isnot possible at the moment orin the near future.

y Porous medium approach isused so that the pebble bed isdivided into bigger cells whereparameters are averaged.

y Temperatures in calculationcells are written as the result.

Figure: Calculation mesh of the T-Hmodel.


Coupling codey A coupling code is written in

PERL.

y Creates input files.

y Controls the calculationprocesses.

y Determines convergence.

Figure: Coupled calculation scheme


Remaining obstacles anddevelopment areas

y Memory requirements of thematerial data and thecalculation power demand.

y Fundamental differencesbetween the codes (discretevs. continuum).

y T-H model needs more work.

y Mapping the packingstructure accurately to theT-H model cells.

y Developing the heat transfermodel.

y Finalizing the coupling code.

Figure: Serpent model showing differenttemperature regions after a T-H iteration


Test casey A test case with 50,000 pebbles, 30 different

temperatures for the pebbles and 30 · 106 neutronhistories was calculated.

y Calculations were done with an Intel Xeon E5520,2.27GHz, 12 GB utilizing 4 parallel MPI tasks.

y Five iteration cycles took about 30 hours.


Conclusions


Conclusionsy Serpent is more than well equipped for PBR

calculations.

y Complicated geometries can be built and thedouble-heterogeneous fuel modeled in full detail.

y Serpent is relatively fast in PBR calculations.

y Coupling of Serpent with T-H in for PBR calculationscan be done with moderate effort.


Thank you for your attention!


pebble bed reactor modeling using...

Documents