CN110532093A - Parallel task division method for multi-geometric-shape full core sub-channels of numerical nuclear reactor - Google Patents
Parallel task division method for multi-geometric-shape full core sub-channels of numerical nuclear reactor Download PDFInfo
- Publication number
- CN110532093A CN110532093A CN201910781296.8A CN201910781296A CN110532093A CN 110532093 A CN110532093 A CN 110532093A CN 201910781296 A CN201910781296 A CN 201910781296A CN 110532093 A CN110532093 A CN 110532093A
- Authority
- CN
- China
- Prior art keywords
- subchannel
- subchannels
- subgraph
- nuclear reactor
- parallel task
- 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
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
- G06F9/5027—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
- G06F9/5027—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
- G06F9/505—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals considering the load
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Monitoring And Testing Of Nuclear Reactors (AREA)
Abstract
The invention relates to a parallel task dividing method for multi-geometric-shape full core subchannels of a numerical nuclear reactor, which is characterized in that a global index is established for components, fuel rods and subchannels in a reactor core, unique numbers are given, the corresponding relation between the components, the fuel rods and the subchannels is established, then the connection relation between the subchannels and the subchannels is mapped into an undirected graph, the subchannels are used as basic objects, the global subchannels are divided into all calculation units, the subchannels divided onto each calculation unit form divided domains, and a ghost subchannel region is established between the spatially adjacent divided domains. The method converts the sub-channel parallel task division problem into a graph division problem, can enable parallel thermal hydraulic simulation software to run by using any number of MPI processes through a self-adaptive region division algorithm, and can ensure load balance and communication minimization.
Description
Technical field
The invention belongs to the thermal-hydraulic analogue techniques of nuclear reactor, and in particular to a kind of more geometric forms of numerical value nuclear reactor
Shape Whole core subchannel parallel task division methods.
Background technique
In the thermal-hydraulic simulation of nuclear reactor, the region between fuel rod is assumed to be " a son by subchannel method
Channel " is modeled for subchannel, the states such as temperature, flow velocity in sunykatuib analysis subchannel.Since subchannel method needs
Momentum and energy equation are solved outside the subchannel that fuel rod surrounds plus on the control volume of axial stratification and solve pressure coefficient square
Battle array is difficult to meet the needs of this computing capability and amount of ram for Whole core simulation on single machine or small cluster.Therefore,
Carrying out the simulation of Whole core thermal-hydraulic using concurrent technique is an inevitable and important guiding.Generally, we simulate reactor core
Region is divided as unit of subchannel, is assigned on different computing units, and parallel subchannel simulation calculus is carried out.Its
It is an important component in parallel sub-channel simulation program that middle task, which divides, consider component, fuel rod, subchannel,
While mapping relations between dividing domain, it is also necessary to the problems such as emphasis considers the load balancing of division result, influences on communication.
Simultaneously as the geometry of heap-type not of the same race is also not quite similar to each other, divided and area maps also band to subchannel task
Not small challenge is carried out.
Summary of the invention
The purpose of the present invention is to provide a kind of numerical value nuclear reactor Whole core subchannel parallel task division methods, are allowed to
It can adapt in the region division of the heap-type of different geometries, also, can also be further ensured that the negative of different computing units
Carry balanced and communication low amounts.
Technical scheme is as follows: a kind of more geometry Whole core subchannel parallel tasks of numerical value nuclear reactor stroke
Divide method, comprising:
(1) global index is established to component, fuel rod and the subchannel in reactor core, gives unique number, and establish group
Corresponding relationship between part-fuel rod-subchannel；
(2) the connection relationship mapping between subchannel and subchannel is become into a non-directed graph；
(3) non-directed graph is divided, is divided into p subgraph Gi(0≤i < p), corresponding to each subgraph
The set of subchannel constitutes a dividing domain；For each dividing domain, part of dividing domain spatially adjacent thereto is logical
The set in road constitutes the ghost subchannel region of the dividing domain, and division should meet following constraint condition:
For all subgraph Gi(0≤i < p), GiNon-empty；
For all subgraph Gi(0≤i < p), GiConnection；
It (4) will be in the corresponding subchannel assignment of each dividing domain a to computing unit.
Further, the more geometry Whole core subchannel parallel task division methods of numerical value nuclear reactor as described above,
Corresponding relationship between component-fuel rod-subchannel described in step (1) includes: the fuel rod for including in various components, which
The subchannel that several fuel rods surround.
Further, the more geometry Whole core subchannel parallel task division methods of numerical value nuclear reactor as described above,
In the non-directed graph described in step (2), subchannel is expressed as the node in figure, and such as two subchannels are spatially adjacent, then exists
The a line for connecting corresponding two nodes is added in figure.
Further, the more geometry Whole core subchannel parallel task division methods of numerical value nuclear reactor as described above,
When dividing in step (3) to the non-directed graph, following two principles should be taken into account:
A. each subgraph to be bordered by subgraph quantity few as far as possible；
B. each subgraph GiNodal point number load in (0≤i < p) is balanced as far as possible.
Further, the more geometry Whole core subchannel parallel task division methods of numerical value nuclear reactor as described above,
The generation method of ghost subchannel described in step (3) is as follows: the subgraph for being first depending on division establishes cutting for each dividing domain
Side collection, it is for each side that cut edge is concentrated, the node join for belonging to the dividing domain in two connected nodes of the side is adjacent
Dividing domain is occupied as its ghost node, corresponding to the ghost subchannel in subchannel division, and records the ghost subchannel category
In dividing domain.
Beneficial effects of the present invention are as follows: the more geometry Whole core subchannels of numerical value nuclear reactor provided by the invention are simultaneously
Row task division methods can make parallel thermal-hydraulic simulation softward use any by adaptive region partitioning algorithm
The MPI process of number is run, and can proof load equilibrium with communicate minimum.The processing mode that the present invention is divided by figure,
So that subchannel divides, the only partition problem of node in figure and side unrelated with reactor core shape, to be adapted to different heap-type
Geometry.
Detailed description of the invention
Fig. 1 is that the connection relationship mapping between subchannel is become non-directed graph (part, solid wire frame in the embodiment of the present invention
In part) in node and side relation schematic diagram；
Fig. 2 is that one kind that 44 subchannels are divided into 4 dividing domains (labeled as 1,2,3,4) in the embodiment of the present invention can
The figure division result of energy；
Fig. 3 establishes the schematic diagram of ghost subchannel for a dividing domain in the embodiment of the present invention for Fig. 2.
Specific embodiment
The present invention is described in detail below with reference to the accompanying drawings and embodiments.
Present invention introduces adaptive subchannel parallel regions division methods, can adapt to any number of MPI process number
(theoretically, minimum MPI process number is 1, and most MPI process numbers are Whole core number of subchannels).In the operation of region division program, need
Read relevant configuration information (such as components distribution, related geometric information, power distribution)；It is right before carrying out region division
Component, fuel rod and subchannel in reactor core establish global index, give unique number, and it is logical to establish component-fuel rod-son
Corresponding relationship between road (surrounds a subchannel and (wherein, surrounds in a component comprising more fuel rods, several fuel rods
The fuel rod of subchannel can be distributed on different components)).It should be pointed out that by defining relevant interface in partition program,
So that the heap-type of different geometries accesses into which can be convenient, it can quickly and easily realize that component-fuel rod-son is logical
The calculating of the geological information in road and the corresponding relationship between it.
After completion number is corresponding with component-fuel rod-subchannel relationship, using subchannel as basic object, by the overall situation
Subchannel is divided on each computing unit, and the subchannel being divided on each computing unit forms dividing domain.Due to coolant
Mobility, the calculating of a subchannel depends on several surrounding neighbours' subchannels, and dividing domain boundary subchannel
Neighbours' subchannel be likely located in neighboring process, so before the computation, needing to obtain to be located in neighboring process has dependence
Subchannel, so that when subsequent calculating only related data need to be obtained from local process.Therefore, it for convenience of parallel communications, avoids repeatedly
It is communicated with neighboring process, it is also necessary to ghost subchannel region is established between spatially adjacent dividing domain.Calculate it
Before, first there is the subchannel positioned at neighboring process of dependence disposably to obtain, when subsequent calculating, it need to only obtain and lead in local process
Road.Finally, the corresponding relationship between division result and component-fuel rod-subchannel-dividing domain is written to output.
In above-mentioned steps, core content is to divide subchannel to dividing domain and establish ghost subchannel region.
Method provided by the present invention establish component-fuel rod-subchannel index with after corresponding relationship, by subchannel with
Connection relationship mapping between subchannel becomes a non-directed graph, can refer to shown in Fig. 1.Square in Fig. 1, where non-zero number
4 vertex positions of shape are fuel rod, and the intermediate region that multiple fuel rods surround is subchannel, and subchannel is expressed as node, such as
Two subchannels spatially have syntople, then a line for connecting corresponding two nodes is added in figure.It is reflected by this
Mode is penetrated, a Graph partition problem subchannel parallel task partition problem can be converted to.
So, it is exactly that non-directed graph G is divided into p subgraph that subchannel parallel task, which divides, so that it meets following divide
Constraint condition:
1. for all subgraph Gi(0≤i < p), GiNon-empty；
2. for all subgraph Gi(0≤i < p), GiConnection.
The quality of the result of division can be measured by following standard:
A. cut edge minimizes (or smallizationer) principle, and such division is so that each subgraph is bordered by subgraph quantity as far as possible
It is few, due to it is parallel when, an a subgraph corresponding process when communication, needs to communicate with each neighboring process and (is bordered by subgraph
Corresponding neighboring process), the number of neighboring process (being bordered by subgraph) influences the number of communication, is bordered by subgraph quantity and less can as far as possible
Guarantee the minimum (or smallizationer) of the traffic in parallel；
B. subgraph GiInterior knot number load balancing principle, the principle are the measurement indexs of computational load equilibrium.
Example shown in Fig. 2 is possible for one kind that 44 subchannels are divided into 4 dividing domains (labeled as 1,2,3,4)
Spatially adjacent subchannel is divided into different dividing domains (i.e. parallel computation unit) by figure division result, dotted line expression,
Subchannel quantity in each dividing domain is respectively 12,12,8,12.
Fig. 3 is the example for establishing ghost subchannel for a dividing domain of Fig. 2.Wherein, black circle is to should belong to this
The subchannel of dividing domain, soft dot corresponds to ghost subchannel, and (such as 15 work song channels belong to No. 2 dividing domains, 16,17 work songs
Channel belongs to No. 3 dividing domains, and 18 work song channels belong to No. 4 dividing domains).
The operating procedure of subchannel parallel regions provided by the invention division methods is detailed below:
1) initialization operation: reading configuration file, initializes reactor core geological information and power distribution；
2) respectively to component, the unique id number of fuel rod, and the corresponding relationship between fuel rod-component is established, i.e., often
Which component a piece fuel rod belongs to；
3) corresponding relationship and subchannel between the subchannel and fuel rod that setting fuel rod surrounds, and it is logical to give each height
The unique id number in road；
4) it initializes subchannel non-directed graph: traversing all subchannels, be added in figure and be used as node, in traversing graph
All nodes, the subchannel for judging whether and being newly added is spatially adjacent, if adjacent, establishes a line with the two nodes；
5) parallel task division (domain division) is carried out: using Kernighan/Lin, Spectral Bisection etc.
Algorithm or figure divide software (such as metis) and carry out side division to figure, and the side for setting division integrates as Es；Each subpicture tag is one
Dividing domain；The subgraph that is bordered by that load balancing principle and subgraph are taken into account when division lacks principle as far as possible, can be by way of weight calculation
Determine final division result.For example, selection k=w1*L+w2*G, wherein L is load imbalance degree, and G is the neighbour of all subgraphs
The sum of subgraph number is connect, w1 and w2 are the weight of two kinds of factors respectively, and by comparing the value of k, smaller is as final division side
Case it is selected (if k value is identical, then it is assumed that one of arbitrarily all can).
6) the ghost subchannel region of dividing domain is established: if there is a line e ∈ Es(the cut edge collection that Es is the subgraph), then
The node join neighbours dividing domain for belonging to the dividing domain in two nodes that the side is connected is corresponding as its ghost node
Ghost subchannel in subchannel division, and record the dividing domain id that the ghost subchannel belongs to；
7) by component-fuel rod-subchannel correspondence mappings relationship and parallel task division result, that is, ghost subchannel region
It is exported.
Obviously, various changes and modifications can be made to the invention without departing from essence of the invention by those skilled in the art
Mind and range.If in this way, belonging to the model of the claims in the present invention and its equivalent technology to these modifications and changes of the present invention
Within enclosing, then the present invention is also intended to include these modifications and variations.
Claims (5)
1. a kind of more geometry Whole core subchannel parallel task division methods of numerical value nuclear reactor, comprising:
(1) global index is established to component, fuel rod and the subchannel in reactor core, gives unique number, and establish component-combustion
Corresponding relationship between charge bar-subchannel；
(2) the connection relationship mapping between subchannel and subchannel is become into a non-directed graph；
(3) non-directed graph is divided, is divided into p subgraph Gi(0≤i < p), son corresponding to each subgraph are logical
The set in road constitutes a dividing domain；For each dividing domain, the part subchannel of dividing domain spatially adjacent thereto
Set constitutes the ghost subchannel region of the dividing domain, and division should meet following constraint condition:
For all subgraph Gi(0≤i < p), GiNon-empty；
For all subgraph Gi(0≤i < p), GiConnection；
It (4) will be in the corresponding subchannel assignment of each dividing domain a to computing unit.
2. the more geometry Whole core subchannel parallel task division methods of numerical value nuclear reactor as described in claim 1,
Be characterized in that: the corresponding relationship between component-fuel rod-subchannel described in step (1), which includes: in various components, includes
Fuel rod, the subchannel which fuel rod surrounds.
3. the more geometry Whole core subchannel parallel task division methods of numerical value nuclear reactor as described in claim 1,
Be characterized in that: in the non-directed graph described in step (2), subchannel is expressed as the node in figure, and such as two subchannels are spatially
It is adjacent, then a line for connecting corresponding two nodes is added in figure.
4. the more geometry Whole core subchannel parallel task division methods of numerical value nuclear reactor as claimed in claim 3,
It is characterized in that: when dividing in step (3) to the non-directed graph, following two principles should be taken into account:
A. each subgraph to be bordered by subgraph quantity few as far as possible；
B. each subgraph GiNodal point number load in (0≤i < p) is balanced as far as possible.
5. the more geometry Whole core subchannel parallel task division methods of numerical value nuclear reactor as described in claim 3 or 4,
It is characterized by: the generation method of ghost subchannel described in step (3) is as follows: the subgraph for being first depending on division is established respectively
The cut edge collection of a dividing domain will belong to the dividing domain in two connected nodes of the side for each side that cut edge is concentrated
Node join neighbours dividing domain as its ghost node, corresponding to subchannel divide in ghost subchannel, and recording should
The dividing domain that ghost subchannel belongs to.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910781296.8A CN110532093B (en) | 2019-08-23 | 2019-08-23 | Parallel task division method for multi-geometric-shape full core sub-channels of numerical nuclear reactor |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910781296.8A CN110532093B (en) | 2019-08-23 | 2019-08-23 | Parallel task division method for multi-geometric-shape full core sub-channels of numerical nuclear reactor |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110532093A true CN110532093A (en) | 2019-12-03 |
| CN110532093B CN110532093B (en) | 2022-05-13 |
Family
ID=68662596
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910781296.8A Active CN110532093B (en) | 2019-08-23 | 2019-08-23 | Parallel task division method for multi-geometric-shape full core sub-channels of numerical nuclear reactor |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110532093B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111008070A (en) * | 2019-12-10 | 2020-04-14 | 北京科技大学 | Fast neutron reactor full reactor core subchannel parallel task division method and system |
| CN111159870A (en) * | 2019-12-20 | 2020-05-15 | 北京科技大学 | Fast reactor isomorphic assembly whole reactor core rod bundle-subchannel mapping construction method and system |
| CN111159865A (en) * | 2019-12-18 | 2020-05-15 | 北京科技大学 | Full-core thermal hydraulic subchannel simulation method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101385090A (en) * | 2006-02-18 | 2009-03-11 | 阿利发Np有限公司 | Fuel assembly for a pressurized water nuclear reactor |
| US20140254737A1 (en) * | 2013-03-06 | 2014-09-11 | Nuscale Power, Llc | Managing nuclear reactor spent fuel rods |
| CN104765589A (en) * | 2014-01-02 | 2015-07-08 | 广州中国科学院软件应用技术研究所 | Grid parallel preprocessing method based on MPI |
-
2019
- 2019-08-23 CN CN201910781296.8A patent/CN110532093B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101385090A (en) * | 2006-02-18 | 2009-03-11 | 阿利发Np有限公司 | Fuel assembly for a pressurized water nuclear reactor |
| US20090122945A1 (en) * | 2006-02-18 | 2009-05-14 | Areva Np Gmbh | Fuel Assembly for a Pressurized-Water Nuclear Reactor |
| US20140254737A1 (en) * | 2013-03-06 | 2014-09-11 | Nuscale Power, Llc | Managing nuclear reactor spent fuel rods |
| CN104765589A (en) * | 2014-01-02 | 2015-07-08 | 广州中国科学院软件应用技术研究所 | Grid parallel preprocessing method based on MPI |
Non-Patent Citations (3)
| Title |
|---|
| MING P: "Parallel analysis of linear systems in corth and kylin2", 《JOURNAL OF NUCLEAR ENGINEERING AND RADIATION SCIENCE》 * |
| 单莹: "基于SMP集群的多层次并行编程模型与并行优化技术", 《计算机应用研究》 * |
| 宋仕钊: "海洋核动力平台堆芯子通道分析", 《原子能科学技术》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111008070A (en) * | 2019-12-10 | 2020-04-14 | 北京科技大学 | Fast neutron reactor full reactor core subchannel parallel task division method and system |
| CN111159865A (en) * | 2019-12-18 | 2020-05-15 | 北京科技大学 | Full-core thermal hydraulic subchannel simulation method |
| CN111159865B (en) * | 2019-12-18 | 2021-10-12 | 北京科技大学 | Full-core thermal hydraulic subchannel simulation method |
| CN111159870A (en) * | 2019-12-20 | 2020-05-15 | 北京科技大学 | Fast reactor isomorphic assembly whole reactor core rod bundle-subchannel mapping construction method and system |
| CN111159870B (en) * | 2019-12-20 | 2021-12-03 | 北京科技大学 | Fast reactor isomorphic assembly whole reactor core rod bundle-subchannel mapping construction method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110532093B (en) | 2022-05-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110532093A (en) | Parallel task division method for multi-geometric-shape full core sub-channels of numerical nuclear reactor | |
| CN108833144B (en) | Intelligent ammeter concentrator site selection method based on cluster statistical model | |
| CN106683171A (en) | GPU multi-thread texture mapping SystemC modeling structure | |
| CN108799844B (en) | Fuzzy set-based water supply network pressure monitoring point site selection method | |
| CN107908913B (en) | Earth power digital-analog method based on parallel computer | |
| CN107291539B (en) | Cluster program scheduler method based on resource significance level | |
| CN104765701B (en) | Data access method and equipment | |
| CN108536692A (en) | A kind of generation method of executive plan, device and database server | |
| CN102857560A (en) | Multi-service application orientated cloud storage data distribution method | |
| CN108520142A (en) | A kind of group of cities Boundary Recognition method, apparatus, equipment and storage medium | |
| CN111198918A (en) | Data processing system based on big data platform and link optimization method | |
| CN104615835B (en) | A kind of engine intercooler analysis method | |
| CN108921296A (en) | A kind of virtual map method and system of quantum bit | |
| CN113821332A (en) | Method, apparatus, device and medium for tuning efficiency of automatic machine learning system | |
| CN103149840A (en) | Semanteme service combination method based on dynamic planning | |
| CN106485030A (en) | A kind of symmetrical border processing method for SPH algorithm | |
| CN102831230B (en) | A kind of parts selection method and system of hybrid power assembly | |
| CN106815320B (en) | Investigation big data visual modeling method and system based on expanded three-dimensional histogram | |
| CN111414725B (en) | Method and device for modeling FPGA (field programmable Gate array) dynamically-expandable software wiring structure | |
| CN109299494B (en) | Data reconstruction method for reactor core thermal hydraulic multi-scale coupling calculation | |
| CN106780747A (en) | A kind of method that Fast Segmentation CFD calculates grid | |
| CN106294540A (en) | Multiple spot geological statistics modeling method based on p stable local sensitivity Hash retrieval Data Styles | |
| CN110275895A (en) | It is a kind of to lack the filling equipment of traffic data, device and method | |
| CN115408543A (en) | Multi-source geological data processing method and system based on three-dimensional geological model | |
| CN115169265A (en) | Method, system, device and medium for analyzing blending coefficient based on numerical analysis |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |