CN101861609A - Reactor dosimetry applications using a parallel 3-D radiation transport code - Google Patents

Reactor dosimetry applications using a parallel 3-D radiation transport code Download PDF

Info

Publication number
CN101861609A
CN101861609A CN200880116643A CN200880116643A CN101861609A CN 101861609 A CN101861609 A CN 101861609A CN 200880116643 A CN200880116643 A CN 200880116643A CN 200880116643 A CN200880116643 A CN 200880116643A CN 101861609 A CN101861609 A CN 101861609A
Authority
CN
China
Prior art keywords
territory
radiation field
field distribution
processor
radiation
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN200880116643A
Other languages
Chinese (zh)
Other versions
CN101861609B (en
Inventor
G·郎格尼
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
CBS Corp
Original Assignee
Westinghouse Electric Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Westinghouse Electric Corp filed Critical Westinghouse Electric Corp
Publication of CN101861609A publication Critical patent/CN101861609A/en
Application granted granted Critical
Publication of CN101861609B publication Critical patent/CN101861609B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • GPHYSICS
    • G09EDUCATION; CRYPTOGRAPHY; DISPLAY; ADVERTISING; SEALS
    • G09BEDUCATIONAL OR DEMONSTRATION APPLIANCES; APPLIANCES FOR TEACHING, OR COMMUNICATING WITH, THE BLIND, DEAF OR MUTE; MODELS; PLANETARIA; GLOBES; MAPS; DIAGRAMS
    • G09B23/00Models for scientific, medical, or mathematical purposes, e.g. full-sized devices for demonstration purposes
    • G09B23/06Models for scientific, medical, or mathematical purposes, e.g. full-sized devices for demonstration purposes for physics
    • GPHYSICS
    • G09EDUCATION; CRYPTOGRAPHY; DISPLAY; ADVERTISING; SEALS
    • G09BEDUCATIONAL OR DEMONSTRATION APPLIANCES; APPLIANCES FOR TEACHING, OR COMMUNICATING WITH, THE BLIND, DEAF OR MUTE; MODELS; PLANETARIA; GLOBES; MAPS; DIAGRAMS
    • G09B19/00Teaching not covered by other main groups of this subclass
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00Computer-aided design [CAD]
    • G06F30/20Design optimisation, verification or simulation
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for program control, e.g. control units
    • G06F9/06Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/44Arrangements for executing specific programs
    • G06F9/445Program loading or initiating
    • G06F9/44521Dynamic linking or loading; Link editing at or after load time, e.g. Java class loading
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for program control, e.g. control units
    • G06F9/06Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/44Arrangements for executing specific programs
    • G06F9/455Emulation; Interpretation; Software simulation, e.g. virtualisation or emulation of application or operating system execution engines
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for program control, e.g. control units
    • G06F9/06Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46Multiprogramming arrangements
    • G06F9/54Interprogram communication
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00Details relating to CAD techniques
    • G06F2111/10Numerical modelling

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Software Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Educational Technology (AREA)
  • Educational Administration (AREA)
  • Algebra (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Computational Mathematics (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Mathematical Physics (AREA)
  • Pure & Applied Mathematics (AREA)
  • Computer Hardware Design (AREA)
  • Evolutionary Computation (AREA)
  • Geometry (AREA)
  • Monitoring And Testing Of Nuclear Reactors (AREA)
  • Measurement Of Radiation (AREA)

Abstract

The invention relates generally to a method for the calculation of radiation field distributions employing a new parallel 3-D radiation transport code and, a multi-processor computer architecture. The code solves algorithms using a domain decomposition approach. For example, angular and spatial domains can be partitioned into subsets and, the subsets can be independently allocated and processed.

Description

Use the reactor dosimetry of parallel 3-D radiation transport code to use
The cross reference of related application
The right of priority that No. 60/975,525, the U.S. Provisional Patent Application of this temporary patent application requirement submission on September 27th, 2007.
Technical field
The calculating of relate generally to radiation field distribution of the present invention, and particularly useful in the neutron dosimetry response (neutron-dosimetry response) of prediction to nuclear reactor chamber and internal part.
Background technology
Can make ins all sorts of ways obtains to transport for neutron and gamma radiation the numerical solution of the linear Boltzmann equation (LBE) of application.Discrete vertical mark method (S N) be a kind of such method that is used in particular for the nuclear engineering field.S NThe numerical solution of equation obtains by the concurrent discretize (concurrent discretization) of phase space (that is, angle domain, spatial domain and energy territory).The concurrent discretize of phase space causes S NTherefore a large amount of unknown quantitys in the equation, need a large amount of computational resources to solve this problem.
Transport application for big 3-D neutron and gamma, use S NThe needed primary memory of numerical solution that equation produces LBE may surpass the current computing power of typical uniprocessor workstation.For example, at being characterized as near 1,500,000 space lattices (spatial mesh), S 8The P of quadrature group (quadrature set), scattering nucleus 3Launch and the complete 3-D neutron transport problem of the typical 2-loop pressurized water reactor (PWR) of 47-neutron energy group (energy group) find the solution the main memory requirements that can cause near the 45G byte.Required a large amount of computational resources may hinder and use the uniprocessor workstation to find the solution such problem.
Expectation is by developing S NThe new derivation algorithm of equation overcomes these difficulties with the advantage of utilizing multiprocessor counting system structure (being distributed memory architecture).For example, expectation structure is linked at together a plurality of workstations independently physically via network backbone (network backbone), is commonly referred to the computing environment of cluster computing environment with foundation.This computing platform had obtained to use widely in the last few years, particularly calculated and extensive numerical simulation field in science.Yet, be necessary the algorithm of design specialized, so that develop the ability of cluster environment.
Therefore, at S NThere is the space of improving in the set of the derivation algorithm of equation with the advantage of utilizing multiprocessor counting system structure.For the calculating such as the radiation field distribution of neutron and gamma radiation field distribution, also there is room for improvement in the method that obtains the numerical solution of LBE.In addition, predict that in mode accurately and efficiently there is room for improvement in the method to the dosimetry response of using in the nuclear reactor.
Summary of the invention
A kind of method that is used to calculate radiation field distribution, comprise: use 3-D radiation transport computer code, described code comprises the territory decomposition algorithm that comprises the territory of selecting from the group of being made up of angle domain and spatial domain, wherein, described territory is distributed independently on the multiprocessor computer architecture and is handled.
A kind of computer program that is used to calculate radiation field distribution.This program comprises code segment, and this code segment is divided (partition) with angle domain and spatial domain when being performed be subclass, distributes each subclass independently, and on the multiprocessor architecture each subclass of processing.
Description of drawings
Fig. 1 a illustrates the geometrical construction and the distribution of material of the 3-D Transport Model of 2-loop PWR; Fig. 1 b illustrates the 2-D part of this model on the x-y plane at z=0.0 place of 2-loop PWR.
Fig. 2 illustrates the measured value and calculated value (M/C) ratio of the dosimetry data of using directed θ weighting (Directional Theta Weighted) adaptive differential computation schemes.
Position (3c) did not proofread and correct the M/C ratio that thickness is compared at the bottom of Fig. 3 a, Fig. 3 b and Fig. 3 c illustrated and have the correction reactor pressure vessel of proofreading and correct thickness and reactor core (core) pushes up position (3a), reactor core midplane (mid-plane) position (3b) and reactor core.
Fig. 4 illustrates the speedup of using different territory decomposition strategies to obtain at a series of processors (for example, reaching 20 processors).
Fig. 5 illustrates the process flow diagram of the embodiment of the invention, wherein, realizes 3-D radiation transport computer program, to produce radiation field distribution.
Embodiment
The present invention relates to the computing method of the radiation field distribution of system.Radiation field can comprise neutron and gamma radiation.In one aspect of the invention, can calculate the outer neutron dosimetry response of container in the nuclear reactor chamber.The kind of nuclear reactor is unrestricted, and can comprise various commercial designs as known in the art.Suitable reactor can include but not limited to: pressurized water reactor (PWR) and boiling water reactor (BWR).For illustrative ease, will this aspect of the present invention be described with reference to the commercial PWR of 2-loop.In the design of PWR and operation, use the microcomputer modelling of the PWR that produces the dosimetry response.
Method of the present invention comprises the application of the parallel radiation transport code of 3-D, and the parallel radiation transport code of 3-D is called RAPTOR-M3G (radiation fast parallel transport the geometrical construction of 3D more than) here.Transport code is provided for finding the solution S NThe set of the parallel algorithm of equation.This method is based on the territory decomposition algorithm, and wherein, spatial domain, angle domain and/or energy territory are divided into can be by the subclass of distributing independently and handling on the multiprocessor architecture.The example of the parallel determinacy transport code of known suitable 3-D comprises PENTRAN TMAnd PARTISN (Sjoden G.E. and Haghighat A., " PENTRAN-Parallel EnvironmentNeutralparticle TRANport in 3-D Cartesian Geometry; " Proceedingsof the Joint International Conference on Mathematical Methods andSupercomputing for Nuclear Applications, Vol.1, pp.232-234, Saratoga Springs, NY (1997)).Compare with traditional uniprocessor application, method of the present invention has reduced the computational load and the storage requirement of each processor, for big 3-D problem provides method for solving efficiently.
Use the parallel storehouse of message passing interface (MPI) in Fortran 90, to develop the RAPTOR-M3G computer code.(Gropp?W.,Lusk?E.,and?Skjellum?A.,Using?MPI?Portable?Parallel?Programming?with?the?Message?PassingInterface,The?MIT?Press,Cambridge,Massachusetts(1999))。The certain characteristics of RAPTOR-M3G comprises following content:
Multigroup (multi-group) S on 3-D Descartes (RAPTOR-XYZ) on the grid of non-homogeneous orthogonal configuration and the cylindrical geometry structure (RAPTOR-RTZ) NEquation separate (M.A.Hunter, G.Longoni, and S.L.Anderson, " Extension of RAPTOR-M3Gto r-θ-z geometry for use in reactor dosimetry applicatons, " Proceedingsof the 13 ThInternatioanl Symposium on Reactor Dosimetry, TheNetherlands (2008));
The space of spatial domain, angle domain and coupling/angle domain decomposition algorithm;
Positive definite weighted difference scheme (positive definite weighted differencingscheme): 0/ θ weighting and directed θ weighting;
Automatic generation (Longoni G.et al. up to the energy level of progression 20 symmetry quadrature group, " Investigation of New Quadrature Sets for Discrete Ordinates Methodwith Application to Non-Conventional Problems; " Transactions of theAmerican Nuclear Society, Vol.84, pp.224-226 (2001));
Parallel storage: allow this locality (local) of spatial sub-regions and angle subdomain to distribute, thereby reduce the storage requirement of each processor;
Parallel allocating task: on multiprocessor to S NEquation is parallel finds the solution, and compares with the uniprocessor technology and has reduced computing time;
Parallel I/O: each its memory device of this accessing of processor, to reduce the I/O time; And
With BOT3P (R.Orsi, " Potential Enhanced Performances inRadiation Transport Analysis on Structured Mesh Grids Made Availableby BOT3P; " Nuclear Science and Engineering, Vol.157, pp.110-116 (2007)), the compatibility and the integration of automatic mesh generator and GIP multigroup cross section pretreater.
Fig. 1 a illustrates the geometrical construction of the 3-D Transport Model of the embodiment of the invention, for example 2-loop PWR.PWR can comprise: 12 feet nuclear reactor cores (nuclear core), thermoshield design and 3 inches reactor cavity spaces.Model geometric structure comprises: reactor core-aqueous mixtures (core-water mixture), reactor core cover (core shroud), reactor core tube (core barrel), thermoshield, comprise reactor pressure vessel (RPV) and the reflective isolating body of stainless steel lining (liner).The RPV of PWR typically has general cylindrical shape, and for example by bottom (head) (bottom head) and removable cover head (top head) in closed at both ends.Use steel-aqueous mixtures that modeling is carried out in above the reactor core and zone, following inside, upper strata (upperinternal) and lower interior part zone.The lower interior part of RPV comprises reactor core tube (that is reactor core supporting construction).This reactor core tube is surrounded by the thermoshield between the inwall of reactor core tube and RPV.In some cases, subpad replaces thermoshield in the use.The reactor core cover is arranged on reactor core tube inside.The annular downcomer is around the reactor core tube.Liquid coolant (generally being water) is recycled in the downcomer.
Fig. 1 b illustrates the 2-D cross section of this model on the x-y plane at z=0.0cm place of 2-loop PWR.Distribution of material among the PWR is shown in addition.Use 5.2 versions of BOT3P code to produce model geometric structure and grid discretization.This model extends to 245.0cm along x axle and y axle from 0.0cm, and extends to 200.0cm along the z axle from-200.0cm.In whole model, use uniform grid; Along the sizing grid that x axle, y axle and z axle are specified 2.0 * 2.0 * 4.0cm respectively, obtain 1,464,100 grids altogether.
Use BUGLE-96 cross-section library (RSICC Data Library CollectionBUGLE-96, " Coupled 47Neutron; 20Gamma-Ray Group CrossSection Library Derived from ENDF/B-VI for LWR Shielding andPressure Vessel Dosimetry Applications; " Oak Ridge NationalLaboratory, Oak Ridge, TN (1999)) and GIP computer code (be DOORS bag a part) (RSICC Computer Code Collection DOORS 3.2a, " One-; Two-and Three-Dimensional Discrete Ordinates Neutron/PhotonTransport Code System; " Oak Ridge National Laboratory, OakRidge, TN (2003)) cross section of handling material blends in the Transport Model.S 8The P of energy level symmetry quadrature group and scattering nucleus 3Spheric harmonic expansion is used to Transport Calculation.The system of passive neutron detector can be installed in the reactor cavity space between reflective isolating body and the pressure vessel.Dosimetry system can provide the accurate information about the fast-neutron irradiation on the belt line region of reactor vessel.The simple metal paper tinsel can be installed in the reactor cavity, and in the aluminum hull of packing into, it makes and produces the distortion minimum of fast neutron spectrum free field effectively and measure.Do not define the neutron dosimetry device that is installed in the reactor cavity space in the Transport Model clearly.
An aspect of of the present present invention comprises and is used for S NThe territory decomposition algorithm of the discretize of equation, this algorithm uses such method: wherein, dimensional energy territory and/or angle energy territory are divided into each subclass that can be distributed independently and handle on the multiprocessor architecture.
S NThe 3D Descartes XYZ version special use that the space of equation and angular discretization and angle domain decomposition algorithm described herein are codes.Because the existence during the scattering redistribution is the S of RAPTOR-RTZ exploitation NThe equation of equation is different with RAPTOR-XYZ.
S NThe phase space of equation is by discretize, that is, and and angle, space and energy.Therefore, the set of resulting linear algebraic equation is suitable for finding the solution on digital machine.Use the multigroup method to be dispersed in the energy territory and turn to a lot of discrete intervals, that is, g=1...G begins (g=1) with the highest energy particle, finishes (g=G) with the minimum energy particle.Transport equation in the multigroup approximation (that is, LBE) is represented with equation (1).
Ω ^ · ▿ → ψ g ( r → , Ω ^ ) + σ g ( r → ) ψ g ( r → , Ω ^ ) = Σ g ′ = 1 G ∫ 4 x d Ω ′ σ gg ′ ( r → , Ω ^ ′ · Ω ^ ) ψ g ′ ( r → , Ω ^ ′ )
+ 1 k χ g Σ g ′ = 1 G v σ f , g ′ ( r → ) φ g ′ ( r → ) + q g e ( r → , Ω ^ ) . - - - ( 1 )
Also angle domain is carried out discretize by the finite set of considering direction by using suitable quadrature integration scheme (quadrature integration scheme).Each discrete direction can be visualized as the lip-deep point of the unit ball with relevant surfaces area, and described relevant surfaces area is corresponding with the weight of quadrature scheme on mathematics.The combination of discrete direction and respective weights is called as the quadrature group.Usually, the quadrature group should satisfy a lot of conditions, so that accurately and be determined on mathematics; Can adopt Several Methods, for example, energy level symmetry quadrature group (LQn) and based on the polynomial quadrature group of Legendre (Longoni G.andHaghighat A., " Development of New Quadrature Sets with theOrdinate Splitting Technique, " Proceedings of the ANS InternationalMeeting on Mathematical Methods for Nuclear Applications (M﹠amp; C2001), Salt Lake City, UT, September 9-13,2001, American NuclearSociety, Inc., La Grange Park, IL (2001)).The quadrature group of exploitation and realization is based on the LQn method in RAPTOR-M3G.
Can carry out discretize to space variable with some technology, for example, finite difference method and Finite Element Method.The equation of developing in RAPTOR-M3G is based on finite difference method, and this finite difference method comprises spatial domain is divided into for example computing unit of refined net, wherein, supposes that the cross section in each unit is constant.In 3D Cartesian geometry structure, use the angular flux of equation (2) estimation in the unit center position.
ψ i , j , k , m , g = q i , j , k + | μ m | a i , j , k , m , g Δx ψ in , x + | η m | b i , j , k , m , g Δy ψ in , y + | ξ m | c i , j , k , m , g Δz ψ in , z | μ m | a i , j , k , m , g Δx + | η m | b i , j , k , m , g Δy + | ξ m | c i , j , k , m , g Δz + σ i , j , k - - - ( 2 )
In equation (2), angle and energy dependence are represented by subscript m and g respectively.Item q I, j, kThe expression be positioned at unit center scattering, fission and external source and.Subscript i, j, k represent the unit center value, and weight a I, j, k, m, g, b I, j, k, m, gAnd c I, j, k, m, gBe limited to the scope between 0.5 and 1.0; RAPTOR-M3G use θ weighting (TW), 0 weighting (ZW) or adaptive directionality θ weighting (DTW) differential scheme calculate each weight (B.Petrovic and A.Haghighat, " the New DirectionalTheta-Weighted S that transports during the scanning (transport sweep) NDifferencing Scheme and Its Application to PressureVessel Fluence Calculations " Proceedings of the 1996RadiationProtection and Shielding Topical Meeting; Falmouth; MA; Vol.1, pp.3-10 (1996)).
Separate S by experiencing each direction that begins from the border of Problem Areas NEquation; This solution procedure is also referred to as and transports scanning.Begin to estimate angular flux from boundary condition or from the boundary angle flux that before adjacent cells, calculated at the center cell location definition.Use equation (2) computing unit central angle flux.Use is called the additional relationships of " differential scheme " and calculates the angular flux that leaves (exit) computing unit.
Carry out in being called the iterative process that the source iterates and transport scanning, this source iterates and also is known as fixed point and iterates, or the Richard is gloomy iterates (Richardson iteration).Continue this process, up to satisfying suitable convergence criterion, promptly, two between iterating with (the Adams M.L.and Larsen E.W. under certain cutoff of the relative error on the scalar flux of any standard, " Fast Iterative Methods for Discrete-Ordinates Particle TransportCalculations; " progress in Nuclear Energy, Vol.40, n.1 (2002)).Calculate for radiation shield, this cutoff generally is set to 1.0e -3Or 1.0e -4
The parallel algorithm of developing in RAPTOR-M3G is based on the decomposition of processor network upper angle territory and/or spatial domain.RAPTOR-M3G creates virtual topology, and correspondingly, a plurality of processors are assigned to angle domain and spatial domain, and the number that is assigned to the processor of angle domain and spatial domain is designated as P respectively aAnd P sThe sum of the processor that any decomposition is required is P n=P aP sBased on this information, processor network is mapped on spatial domain and the angle domain, creates virtual topology, and this virtual topology is with the local subdomain of each relational processor to this processor.
Angle domain is pressed octant (octant) and is divided, and wherein, the processor of appointment is assigned to local octant in order on angle domain.The local number of the octant that each processor distributed is given by equation (3).
N loct = 8 P a - - - ( 3 )
At P aN on the individual processor LoctThe local execution transports scanning on the individual octant; The MPI communicator that is used for angle domain is used to the angular flux between each processor is carried out synchronously, and causes reflective boundary condition.
By in order with P sIndividual processor distribution is given some x-y plane, divides spatial domain along the z axle.At P sDivide on the individual processor (that is, km) along the sum of the refined net of z axle; Map array (being kmloc) is used for P is distributed on the x-y plane sIndividual processor.Distribute to P sThe number on the x-y plane of individual processor is arbitrarily; Yet, need satisfy the condition in the equation (4), so that limit and problem geometrical construction consistent spatial decomposition on topology.
Σ i = 1 P s kmloc ( i ) = km - - - ( 4 )
The dirigibility that processor on the spatial domain is mapped to the x-y plane of arbitrary number can be depended on such fact: the number of z-plane may not be divided exactly by the processor number on the spatial domain.P sThe inhomogeneous division on the x-y plane on the individual processor may cause processor load unbalanced, thereby causes the loss of performance.In the present invention, can application mix angle/spatial decomposition strategy to overcome this difficulty.Mixed decomposition comprises the combination of angle domain and spatial domain, to comprise the concurrent division in these territories.Further specifying hybrid domain in the example 1 below decomposes.
Fig. 5 illustrates the process flow diagram of the embodiment of the invention, wherein, realizes 3-D radiation transport computer program, to produce radiation field distribution.Present embodiment comprises geometry and the material information of acquisition about wanting system for modeling.This information can obtain from each provenance, such as, but not limited to nuclear reactor figure.Can select to calculate the S of required abundance then NProgression.In addition, P that can selected angle territory (that is, scattering nucleus and angular flux) nLaunch progression.Can also select suitable differential scheme (that is, TW, ZW or DTW).Can use the suitable grid generator of Descartes XYZ or RTZ geometrical construction to produce 3-D model with the discretize system.By every kind of material is mixed with suitable data set (such as BUGLE-96), be every kind of material production cross section table.Number and corresponding decomposition strategy at the problem selection processor that will solve.Carry out then and calculate.After calculating, can carry out aftertreatment and analysis to the result who is produced.
Being used for RAPTOR-M3G computer code of the present invention provides the solution of accurate and efficient (for example, the computing time of minimizing) of radiation transport problem.In one aspect of the invention, calculate the outer neutron dosimetry response of container at the chamber in the nuclear reactor vessel.Compare with actual measurement, the bat of the fast neutron reaction in the reactor cavity space of the 2-loop PWR that calculates by RAPTOR-M3G is 96%.In addition, on 20 processor computer clusters, use mixing angle/spatial domain decomposition strategy in about 106 minutes clock time, to obtain separating of transport issues.
Though described specific embodiments of the invention in detail, those skilled in the art should understand according to whole instructions of the present disclosure and can develop the various modifications of those details and substitute.For example, Shuo Ming one aspect of the present invention relates to nuclear industry here, particularly nuclear reactor.Yet the present invention can also be used for other application such as the broad range of medical domain.For example, the present invention can be used to be defined as treatment and/or cure cancer and the patient is executed the radiation dose that send (deliver).Therefore, disclosed specified scheme is illustrative for scope of the present invention, rather than restrictive, and scope of the present invention is provided by the four corner of appended claims and all equivalents thereof.
Example
Example 1---RAPTOR-M3G parallel performance is analyzed
The Transport Calculation of discussing in " example " is with running on (that is RAPTOR-M3G execution on EAGLE-1), of 20 processor computer clusters.This cluster is made up of 5 nodes with 2 double-core dual processor AMD Opteron, 64 bit architectures.Available cluster total memory (that is RAM) is the 40G byte; The network interconnection is characterised in that the 1GBit/s (bandwidth of bit/s).Utilize this hardware configuration, RAPTOR-M3G is in whole 3D Transport Calculation of finishing in about 106 minutes on 20 processors 2-loop PWR.Use DTW, TW or ZW differential scheme not to observe significant performance difference.
In addition, set up the parallel performance of simple test problem with code analysis.This test problem by have equally distributed stationary source, form with the case (box) of 50 * 50 * 50cm of the uniform grid discretize of 1cm.Use S with an energy group cross section collection 8Quadrature group and P0 isotropic scatterning.Use wall clock (wall-clock) time, acceleration and parallel efficiency to assess the parallel performance of RAPTOR-M3G.Acceleration and parallel efficiency in equation 5 and 6, have been defined respectively.
S p=T s/T p (5)
η p=S p/N p (6)
Wherein, T sAnd T pBe respectively that uniprocessor calculates and multiprocessor calculates the required wall clock time.N pBe to be used to realize the wall clock time T pThe number of processor.Fig. 4 illustrates for reaching the comparison that 20 processors use the acceleration of different decomposition strategy acquisitions.
The acceleration that utilizes spatial decomposition to obtain reduces gradually along with the increase of processor number.It is believed that the behavior is because the thinner calculating granularity of each processor causes; Along with spatial domain is broken down into littler subdomain, the operation decreased number of each processor, and the call duration time between the processor increases; Therefore cause performance to descend.Network data transmission between each node is the limiting factor of distributed memory architecture normally.Make iterating of the required bigger quantity of problem convergence further make the performance of spatial decomposition strategy reduce.Yet angle domain and spatial domain have been produced better result by the mixed decomposition of concurrent division.It is believed that the behavior is because the more coarse calculating granularity of being introduced by this decomposition causes; And for mixed decomposition, make problem convergence required iterate number unlike spatial decomposition increase so much.
The ratio of the response that the dosimetry response and the RAPTOR-M3G of example 2---measurement calculates
Between dosimetry response of measuring and corresponding prediction, compare with the RAPTOR-M3G acquisition.Use IRDF-2002 dosimetry storehouse (I.Kodeli and A.Trkov, " Validation of the IRDF-2002Dosimetry Library ", NuclearInstruments and Methods in Physics Research Section A:Accelerators, Spectrometers, Detectors and Associated Equipment, Vol.57, Issue 3, pp.664-681 (2007)), to produce the dosimetry that calculates response at listed neutron reaction in the table 1.
With the dosimetry response of the measurement of listed reaction in the table 1 with compare with the response of RAPTOR-M3G calculating.
The neutron reaction that table 1. is measured with dosimetry system
The material reaction
Copper 63Cu (n, α) 60Co
Iron 54Fe (n, p) 54Mn
Nickel 58Ni (n, p) 58Co
Uranium 238U (n, f) 137Cs
Neptunium 237Np (n, f) 137Cs
Use the metal forming of cadmium shield to come the reaction of listing in the meter 1, thus, the thermal component of neutron spectra is suppressed.
4 azimuth positions at the reactor core midplane place in the reactor cavity space (promptly 0 °, 15 °, 30 ° and 45 °) obtain the response of measurement.Because 2-loop PWR reactor is characterised in that the peak value fast neutron influence of 0 ° of position usually,, therefore obtain additional measurement in this position because nuclear fuel and RPV are extremely approaching.Specifically, locate, on the top of movable reactor core and bottom are in axially, obtain to measure at 0 °.At first, find that the dosimetry response calculated crosses prediction consistently and (over-predict) measured data.
Further research discloses: the RPV thickness that uses in Transport Model has confirmed initial discovery less than detect measured RPV thickness during (ISI) in the use of reactor pressure vessel (RPV).New RPV thickness is introduced Transport Model, and the accuracy of the dosimetry data of calculating has improved average 8%.The ratio of the measured value of the dosimetry data of using DTW adaptive differential computation schemes to calculated value (M/C) has been shown among Fig. 2.As shown in Figure 2, in each position and at every kind of dosimetry material, the M/C ratio changes in 10% scope always.Can predict that wherein, the curvature of system becomes more relevant by using non-uniform grid refinement to reduce in the mistake of 30 ° and 45 ° azimuth position 30 ° and these positions, 45 ° of position angles.Average M/C ratio in all dosimetry positions is 0.96.
Fig. 3 a is illustrated in the M/C value that obtains under the situation of the RPV thickness that ISI proofreaies and correct and in the comparison of the value that does not have to obtain under the situation of thickness correction to Fig. 3 c.All dosimetry samples that are positioned at 0 ° of azimuth position are compared and present this comparison.The RPV thickness of the correction that use ISI measures has improved the accuracy of the response of calculating in each place, dosimetry position.Also obtain similar result at 15 °, 30 ° with 45 ° of azimuth positions.
Measurement and reaction rate calculating in each dosimetry position has been shown in table 2, and has used the M/C ratio that utilizes ISI to measure the RPV thickness of being proofreaied and correct.The average M/C ratio of listing in table 2 in the locational reaction of all dosimetries is 0.96.
Table 2. utilize that the DTW differential scheme to measure with the reaction rate of calculating

Claims (20)

1. method that is used to calculate radiation field distribution comprises:
Use 3-D radiation transport computer code, described code comprises the territory decomposition algorithm that comprises the territory of selecting from the group of being made up of angle domain and spatial domain, and wherein, described territory is assigned with and handles on the multiprocessor computer architecture.
2. method according to claim 1, wherein, described radiation field distribution comprises the radiation field of selecting from the group of being made up of neutron irradiation and gamma radiation.
3. method according to claim 1, wherein, described radiation field distribution is at the nuclear reactor chamber.
4. method according to claim 3, wherein, described nuclear reactor is a 2-loop pressurized water reactor.
5. method according to claim 1, wherein, described algorithm is constructed to the linear Boltzmann equation of numerical solution.
6. method according to claim 1, wherein, described multiprocessor computer architecture comprises by the network connection chain receives together a plurality of workstations independently physically.
7. method according to claim 1 also comprises: described territory is divided on described multiprocessor architecture by the subclass of distributing independently and handling.
8. method according to claim 5, wherein, the described division in described territory comprises the parallel storage that uses on a plurality of processors.
9. method according to claim 1, wherein, described radiation field distribution comprises the outer neutron dosimetry response of the container in the nuclear reactor chamber.
10. method according to claim 7, wherein, in the described dosimetry response of the textural acquisition of 3D Cartesian geometry.
11. method according to claim 1, wherein, described algorithm is constructed to use parallel algorithm to find the solution the SN equation concurrently.
12. method according to claim 1, wherein, equipment each this accessing of processor of described multiprocessor computer architecture its oneself, that from the group of forming by memory device and local storage, select.
13. method according to claim 1 also comprises: turn to a plurality of discrete intervals with described energy territory is discrete.
14. method according to claim 1 also comprises: described spatial domain is divided into computing unit.
15. method according to claim 14, wherein, the cross section of supposition is constant in each unit.
16. method according to claim 1, wherein, decompose in the territory is that angle domain and spatial domain are by the mixed decomposition of concurrent division.
17. method according to claim 16, wherein, the number of processor can reach 20 and comprise 20.
18. method according to claim 17 wherein, is found the solution problem in less than 2 hours time period.
19. method according to claim 17 wherein, is compared with the value that calculates, and has 90% or bigger accuracy by separating of using that described method derives.
20. a computer program that is used to calculate radiation field distribution, described program comprises code segment, and described code segment is divided into subclass with angle domain and spatial domain when carrying out, distribute described subclass independently; And on the multiprocessor architecture, handle described subclass.
CN2008801166435A 2007-09-27 2008-09-25 Reactor dosimetry applications using parallel 3-D radiation transport code Active CN101861609B (en)

Applications Claiming Priority (5)

Application Number Priority Date Filing Date Title
US97552507P 2007-09-27 2007-09-27
US60/975,525 2007-09-27
US5678508P 2008-05-28 2008-05-28
US61/056,785 2008-05-28
PCT/US2008/077625 WO2009042747A1 (en) 2007-09-27 2008-09-25 Reactor dosimetry applications using a parallel 3-d radiation transport code

Publications (2)

Publication Number Publication Date
CN101861609A true CN101861609A (en) 2010-10-13
CN101861609B CN101861609B (en) 2013-12-18

Family

ID=40511847

Family Applications (1)

Application Number Title Priority Date Filing Date
CN2008801166435A Active CN101861609B (en) 2007-09-27 2008-09-25 Reactor dosimetry applications using parallel 3-D radiation transport code

Country Status (5)

Country Link
EP (1) EP2215617A4 (en)
JP (1) JP5460601B2 (en)
KR (1) KR101545035B1 (en)
CN (1) CN101861609B (en)
WO (1) WO2009042747A1 (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106126929A (en) * 2016-06-24 2016-11-16 西安交通大学 The method processing extensive interior vacuum particle transport issues based on DISORT method

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP5289542B2 (en) * 2011-11-25 2013-09-11 日立Geニュークリア・エナジー株式会社 Air dose evaluation apparatus and method
JP2013186016A (en) * 2012-03-09 2013-09-19 Hitachi-Ge Nuclear Energy Ltd Method of measuring three-dimensional space distribution of radioactive ray data
CN104678424B (en) * 2013-12-03 2017-08-25 中广核(北京)仿真技术有限公司 The neutron counting system and method for nuclear power plant analog machine
US11060987B2 (en) 2019-04-23 2021-07-13 Canon Medical Systems Corporation Method and apparatus for fast scatter simulation and correction in computed tomography (CT)
CN110502813B (en) * 2019-08-09 2023-04-18 中国舰船研究设计中心 Nuclear power ship reactor cabin mixed radiation field calculation method based on radiation source discretization
KR102374327B1 (en) * 2019-10-08 2022-03-15 주식회사엔에스이 A 3D System for Estimating Radioactivity of Equipment Elements and Representing Radiation Maps in Nuclear Power Plants
CN112346873B (en) * 2020-11-26 2022-02-11 中国核动力研究设计院 Characteristic line method multistage parallel method suitable for hardware architecture of modern supercomputer

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4666662A (en) * 1984-07-10 1987-05-19 Westinghouse Electric Corp. Steam generator recirculating system for a pressurized water nuclear reactor
US5267276A (en) * 1991-11-12 1993-11-30 The University Of Chicago Neutron transport analysis for nuclear reactor design
US7233888B2 (en) * 2002-07-09 2007-06-19 General Electric Company Monte Carlo criticality-mode systems and methods for computing neutron and gamma fluence in a nuclear reactor
TWI274354B (en) * 2002-09-03 2007-02-21 Gen Electric Methods and apparatus for determining neutron flux in a nuclear reactor
US8109766B2 (en) * 2003-10-03 2012-02-07 Global Nuclear Fuel-Americas Llc Method for predicted reactor simulation
WO2006138513A1 (en) * 2005-06-16 2006-12-28 Nomos Corporation Variance reduction simulation system, program product, and related methods
JP2007278991A (en) * 2006-04-11 2007-10-25 Nuclear Fuel Ind Ltd Control method for three-dimensional mesh system for parallel calculation

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106126929A (en) * 2016-06-24 2016-11-16 西安交通大学 The method processing extensive interior vacuum particle transport issues based on DISORT method
CN106126929B (en) * 2016-06-24 2018-10-19 西安交通大学 Method based on the extensive interior vacuum particle transport issues of DISORT method processing

Also Published As

Publication number Publication date
CN101861609B (en) 2013-12-18
EP2215617A1 (en) 2010-08-11
JP2010540936A (en) 2010-12-24
EP2215617A4 (en) 2014-11-19
KR101545035B1 (en) 2015-08-17
WO2009042747A1 (en) 2009-04-02
KR20100066572A (en) 2010-06-17
JP5460601B2 (en) 2014-04-02

Similar Documents

Publication Publication Date Title
CN101861609B (en) Reactor dosimetry applications using parallel 3-D radiation transport code
Harry et al. Stochastic template placement algorithm for gravitational wave data analysis
Shibata et al. Coalescence of black hole-neutron star binaries
CN106126925A (en) A kind of method improving reactor core three-dimensional netron-flux density FINE DISTRIBUTION
Ryu et al. nTRACER whole core transport solutions to C5G7-TD benchmark
Al Falahi et al. Experimental investigation of the pebble bed structure by using gamma ray tomography
Yadav et al. Monte Carlo real coded genetic algorithm (MC‐RGA) for radioactive particle tracking (RPT) experimentation
Ilas et al. A Monte Carlo based nodal diffusion model for criticality analysis of spent fuel storage lattices
Assogba et al. Precise 3D reactor core calculation using Spherical Harmonics and Discontinuous Galerkin Finite Element Methods
Becker et al. Hydrodynamic instability and dynamic burnout in natural circulation two-phase flow. An experimental and theoretical study
Yunos et al. Reconstruction algorithm of calibration map for RPT techniques in quadrilateral bubble column reactor using MCNPX code
Alberti Steady State Modeling of the Minimum Critical Core of the Transient Reactor Test Facility
US7606686B2 (en) Reactor dosimetry applications using a parallel 3-D radiation transport code
Longoni et al. Reactor dosimetry applications using RAPTOR-M3G: a new parallel 3-D radiation transport code
Gurevich et al. MCU code implementation of the calculation of Functionals in volumes of a supplementary geometry
Poveshchenko et al. Method for calculating the neutron diffusion coefficient in a nuclear reactor cell
Mager et al. Verification of the High Fidelity Model of the CROCUS Experimental Reactor
Mirsky et al. Evolution of cooling tower fill
Martz et al. Second Order Tracking Verification for Unstructured Mesh in the MCNP R Code Post Version 6.2. 0
Pevey Multi-objective Optimization of the Fast Neutron Source by Machine Learning
Yesilyurt Advanced Monte Carlo methods for analysis of very high temperature reactors: On-the-fly Doppler broadening and deterministic/Monte Carlo methods
Dučkić et al. Point Kernel Modification Including Support Vector Regression Neutron Buildup Factor Models
Merzari et al. Cardinal: A Lower Length-Scale Simulator for Fluoride Cooled High Temperature Reactors
Chen et al. Physics-oriented optimization strategy for the energy lookup algorithm in continuous energy Monte Carlo neutron transport simulation
Ferrer et al. Particle Conservation and the Linear Source Method of Characteristics for Isotropic and Anisotropic Sources

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant