NAME OR DESIGNATION OF PROGRAM, COMPUTER, DESCRIPTION OF PROGRAM OR FUNCTION, METHOD OF SOLUTION, RESTRICTIONS ON THE COMPLEXITY OF THE PROBLEM, TYPICAL RUNNING TIME, UNUSUAL FEATURES OF THE PROGRAM, RELATED AND AUXILIARY PROGRAMS, STATUS, REFERENCES, MACHINE REQUIREMENTS, LANGUAGE, OPERATING SYSTEM UNDER WHICH PROGRAM IS EXECUTED, OTHER PROGRAMMING OR OPERATING INFORMATION OR RESTRICTIONS, NAME AND ESTABLISHMENT OF AUTHORS, MATERIAL, CATEGORIES

[ top ]

[ top ]

To submit a request, click below on the link of the version you wish to order.
Only liaison officers are authorised to submit online requests. Rules for requesters are
available here.

Program name | Package id | Status | Status date |
---|---|---|---|

FINELM | NEA-0896/03 | Tested | 05-DEC-1990 |

Machines used:

Package ID | Orig. computer | Test computer |
---|---|---|

NEA-0896/03 | Many Computers | DEC VAX 8810 |

[ top ]

3. DESCRIPTION OF PROGRAM OR FUNCTION

FINELM solves multi-group diffusion theory eigenvalue (direct and adjoint) and source problems in 2- and 3-dimensional space. Geometries provided are x-y, x-y-z, r-z, r-theta, and r-theta-z. Triangular and rectangular Lagrangian elements are used. Distinct orders of approximations may be used along each axis for rectangles. For triangular elements, the appro- ximation in the plane must remain constant but can differ from the approximation in z. The coefficient matrices required for the chosen approximations are generated within the code. Both up- and down- scatterimg are provided. Group-dependent internal boundary conditions may also be considered. Albedos may range from zero to unity, both on internal and external boundaries. In addition, for internal boundary conditions, distinct albedos along each coordinate axis may be specified in order to model the total leakage more exactly and to compensate for the shape of the element. A very flexible choice of output point/element/zone flux normalization is available. A restart option has also been provided.

FINELM solves multi-group diffusion theory eigenvalue (direct and adjoint) and source problems in 2- and 3-dimensional space. Geometries provided are x-y, x-y-z, r-z, r-theta, and r-theta-z. Triangular and rectangular Lagrangian elements are used. Distinct orders of approximations may be used along each axis for rectangles. For triangular elements, the appro- ximation in the plane must remain constant but can differ from the approximation in z. The coefficient matrices required for the chosen approximations are generated within the code. Both up- and down- scatterimg are provided. Group-dependent internal boundary conditions may also be considered. Albedos may range from zero to unity, both on internal and external boundaries. In addition, for internal boundary conditions, distinct albedos along each coordinate axis may be specified in order to model the total leakage more exactly and to compensate for the shape of the element. A very flexible choice of output point/element/zone flux normalization is available. A restart option has also been provided.

[ top ]

4. METHOD OF SOLUTION

A group by group direct solution method with Choleski decomposition of the system matrix is used. A series of smaller sub-problems may be defined using incomplete dissections should the in-core computer storage become a problem. The outer iterations are accelerated by the use of Lebedev (a variant of Chebyshev) accelerations in combination with coarse-mesh rebalancing of the space collapsing type.

A group by group direct solution method with Choleski decomposition of the system matrix is used. A series of smaller sub-problems may be defined using incomplete dissections should the in-core computer storage become a problem. The outer iterations are accelerated by the use of Lebedev (a variant of Chebyshev) accelerations in combination with coarse-mesh rebalancing of the space collapsing type.

[ top ]

5. RESTRICTIONS ON THE COMPLEXITY OF THE PROBLEM

The problem dimen- sions are all variable. The total problem size is limited only by on-line random access disc storage. Since most of the data is generated from very simple input, a rectangular mesh of elements is required. In the plane these are either right-triangular and/or rectangular meshes. Along the third axis a rectangular mesh is used. The degree of approximation along each axis need not necessarily be the same. Should the problem become too large to fit into core, the method of dissections may be implemented. In this manner, the problem is divided into a series of smaller linked sub-problems. It is suggested that the number of energy groups remain small as FINELM is a diffusion code. However, no built-in upper limit exists.

The problem dimen- sions are all variable. The total problem size is limited only by on-line random access disc storage. Since most of the data is generated from very simple input, a rectangular mesh of elements is required. In the plane these are either right-triangular and/or rectangular meshes. Along the third axis a rectangular mesh is used. The degree of approximation along each axis need not necessarily be the same. Should the problem become too large to fit into core, the method of dissections may be implemented. In this manner, the problem is divided into a series of smaller linked sub-problems. It is suggested that the number of energy groups remain small as FINELM is a diffusion code. However, no built-in upper limit exists.

[ top ]

6. TYPICAL RUNNING TIME

Running time is a function of the number of nodes, the orders of approximation, the order of space, the number of energy groups, the width of the up- and down-scatter bands, the usage/non-usage of coarse-mesh rebalancing, the usage/non-usage of Lebedev acceleration, and the number and type of dissections.

The 2 energy group, 2-dimensional problem designated test case1 (an IAEA LWR benchmark, having 164 nodes/group, with no dissections and using both Lebedev accelerations and coarse-mesh rebalancing from the 9th iteration, required 17 seconds to model and assemble the input and Choleski system matrices; and 70 seconds to iterate 20 times to an eigenvalue convergence of 4.0E-6.

NEA 896/03: NEA-DB executed the test cases included in the package on a VAX 8810. The following CPU times were required: case 1A: 27 s; case 1B: 3m03s; case 2: 15s; case 3: 12s; case 4A: 3s; case 4B: 3s; case 5: 5s; case 6: 4s; case 7: 3s; case 8: 3s

Running time is a function of the number of nodes, the orders of approximation, the order of space, the number of energy groups, the width of the up- and down-scatter bands, the usage/non-usage of coarse-mesh rebalancing, the usage/non-usage of Lebedev acceleration, and the number and type of dissections.

The 2 energy group, 2-dimensional problem designated test case1 (an IAEA LWR benchmark, having 164 nodes/group, with no dissections and using both Lebedev accelerations and coarse-mesh rebalancing from the 9th iteration, required 17 seconds to model and assemble the input and Choleski system matrices; and 70 seconds to iterate 20 times to an eigenvalue convergence of 4.0E-6.

NEA 896/03: NEA-DB executed the test cases included in the package on a VAX 8810. The following CPU times were required: case 1A: 27 s; case 1B: 3m03s; case 2: 15s; case 3: 12s; case 4A: 3s; case 4B: 3s; case 5: 5s; case 6: 4s; case 7: 3s; case 8: 3s

[ top ]

7. UNUSUAL FEATURES OF THE PROGRAM

- Very simple user-friendly input.

- Distinct approximation along each axis to minimize excessive allocation of nodes where the flux is relatively flat.

- Higher orders of approximation may be chosen.

- Group-dependent internal boundary conditions.

- Flexible choice of re-normalization of results.

- Simple to use re-start option.

- The ability to dissect the program into a sequence of smaller linked problems if fast core limits are reached.

- Very simple user-friendly input.

- Distinct approximation along each axis to minimize excessive allocation of nodes where the flux is relatively flat.

- Higher orders of approximation may be chosen.

- Group-dependent internal boundary conditions.

- Flexible choice of re-normalization of results.

- Simple to use re-start option.

- The ability to dissect the program into a sequence of smaller linked problems if fast core limits are reached.

[ top ]

8. RELATED AND AUXILIARY PROGRAMS

A postprocessor, REFINE, which uses the FINELM output point fluxes, may be used to arbitrarily sub-divide the meshes used by FINELM to produce refined average fluxes for follow-up calculations. Along an axis, the sub-divisions must be uniform, but may be distinct along each axis.

Operating instructions are distributed on the dispatched tape.

A postprocessor, REFINE, which uses the FINELM output point fluxes, may be used to arbitrarily sub-divide the meshes used by FINELM to produce refined average fluxes for follow-up calculations. Along an axis, the sub-divisions must be uniform, but may be distinct along each axis.

Operating instructions are distributed on the dispatched tape.

[ top ]

10. REFERENCES

- D.M. Davierwalla:

"A Finite Element Solution to the Neutron Diffusion Equation in

Two Dimensions"

ISNM 37 (1977)

- D.M. Davierwalla and C.E. Higgs:

Mathematical and Computational Meeting on Advances in Reactor

Computation, 28-31 March l983, Salt Lake City, U. S. A.

Poster Session Kiosk Address F

- Workshop Seminar on Finite Element Multi-Dimensional Diffusion

Codes, 15-16 September 1983, Saclay, France

NEA Newsletter No. 30 (December 1983)

- D.M. Davierwalla:

"A Finite Element Solution to the Neutron Diffusion Equation in

Two Dimensions"

ISNM 37 (1977)

- D.M. Davierwalla and C.E. Higgs:

Mathematical and Computational Meeting on Advances in Reactor

Computation, 28-31 March l983, Salt Lake City, U. S. A.

Poster Session Kiosk Address F

- Workshop Seminar on Finite Element Multi-Dimensional Diffusion

Codes, 15-16 September 1983, Saclay, France

NEA Newsletter No. 30 (December 1983)

NEA-0896/03, included references:

- D.M. Davierwalla:FINELM: A Multigroup Finite Element Diffusion Code.

Part I: X-Y Geometry and Dissections.

EIR - Bericht Nr. 419 (December 1980)

- D.M. Davierwalla:

FINELM: A Multigroup Finite Element Diffusion Code.

Part II: R-Z Geometry and Numerical Accelerations.

EIR - Bericht Nr. 428 (May 1981)

- C.E. Higgs and D.M. Davierwalla:

FINELM: A Multigroup Finite Element Diffusion Code.

Input Description, Program Description and Test Examples.

EIR - Bericht Nr. 442 (June 1981)

- S. Pelloni, C. Higgs and D.M. Davierwalla:

FINELM: A Multigroup Finite Element Diffusion Code.

Part III: R-Theta Geometry and Internal Boundary Conditions.

EIR - Bericht Nr. 459 (April 1982)

- J. Stepanek and D.M. Davierwalla:

Chapters 3 and 5 of Draft FINELM Manual

- Comparison Tables for Two 3-Dim. Cases on CRAY, SUN and VAX

[ top ]

11. MACHINE REQUIREMENTS

- To date we have not exceeded 24 Mbytes of disc storage (for 3-dimensional x-y-z geometry, cubic approximation in the plane, quadratic in Z, 2-group 3-x dissectors, 2-y dissectors, almost 6000 nodes/group).

- A line printer or hard copy terminal with 132 characters/line would be advantageous.

- Restart file may be stored on tape.

- The VAX 11/780 system clock is used.

- To date we have not exceeded 24 Mbytes of disc storage (for 3-dimensional x-y-z geometry, cubic approximation in the plane, quadratic in Z, 2-group 3-x dissectors, 2-y dissectors, almost 6000 nodes/group).

- A line printer or hard copy terminal with 132 characters/line would be advantageous.

- Restart file may be stored on tape.

- The VAX 11/780 system clock is used.

NEA-0896/03

NEA-DB executed the test cases on a VAX 8810 in 184k bytes of main storage.[ top ]

13. OPERATING SYSTEM UNDER WHICH PROGRAM IS EXECUTED

CRAY --> UNICOS

SUN --> SUN-OS (UNIX)

VAX-STATION --> VMS

CDC --> NOS

CRAY --> UNICOS

SUN --> SUN-OS (UNIX)

VAX-STATION --> VMS

CDC --> NOS

NEA-0896/03

VMS V5.3 (VAX 8810).[ top ]

[ top ]

[ top ]

NEA-0896/03

File name | File description | Records |
---|---|---|

NEA0896_03.001 | Information file | 117 |

NEA0896_03.002 | FINELM.UPD Source file | 9978 |

NEA0896_03.003 | FINELM.VAX VAX version source file | 10036 |

NEA0896_03.004 | UPDATE.FOR Subsidiary program | 112 |

NEA0896_03.005 | VAX.JOB VAX test run procedure | 93 |

NEA0896_03.006 | CRAY.JOB CRAY test run procedure | 147 |

NEA0896_03.007 | SUN.JOB SUN test run procedure | 120 |

NEA0896_03.008 | CASE1A.DAT Input data for test case 1A | 31 |

NEA0896_03.009 | CASE1B.DAT Input data for test case 1B | 71 |

NEA0896_03.010 | CASE2.DAT Input data for test case 2 | 26 |

NEA0896_03.011 | CASE3.DAT Input data for test case 3 | 20 |

NEA0896_03.012 | CASE4A.DAT Input data for test case 4A | 22 |

NEA0896_03.013 | CASE4B.DAT Input data for test case 4B | 20 |

NEA0896_03.014 | CASE5.DAT Input data for test case 5 | 17 |

NEA0896_03.015 | CASE6.DAT Input data for test case 6 | 16 |

NEA0896_03.016 | CASE7.DAT Input data for test case 7 | 16 |

NEA0896_03.017 | CASE8.DAT Input data for test case 8 | 16 |

NEA0896_03.018 | XSLIB1.DAT Cross-section library | 21 |

NEA0896_03.019 | XSLIB2.DAT Cross-section library | 36 |

NEA0896_03.020 | XSLIB3.DAT Cross-section library | 51 |

NEA0896_03.021 | XSLIB4A.DAT Cross-section library | 10 |

NEA0896_03.022 | XSLIB4B.DAT Cross-section library | 10 |

NEA0896_03.023 | XSLIB5.DAT Cross-section library | 4 |

NEA0896_03.024 | FINOUT.DAT VAX output obtained at NEADB | 6859 |

NEA0896_03.025 | OUTPUT.XMP CRAY-XMP output | 7575 |

NEA0896_03.026 | OUTPUT.SUN SUN output | 6945 |

Keywords: Lagrange equations, coarse mesh, diffusion, eigenvalues, finite element method, finite elements, multigroup, r-theta, r-z, rectangular, three-dimensional, triangular, two-dimensional, x-y, x-y-z.