Skip to content

Commit b041713

Browse files
committed
FDS User Guide: add section on ULMAT HYPRE
1 parent e26d751 commit b041713

File tree

2 files changed

+69
-60
lines changed

2 files changed

+69
-60
lines changed

Manuals/Bibliography/FDS_general.bib

Lines changed: 66 additions & 59 deletions
Original file line numberDiff line numberDiff line change
@@ -692,14 +692,14 @@ @MASTERSTHESIS{Bittern:Thesis
692692

693693

694694
@TECHREPORT{NRC:VICTORIA,
695-
author = {N.E. Bixler},
696-
title = {{VICTORIA 2.0: A Mechanistic Model for Radionuclide Behavior in a Nuclear Reactor Coolant System Under Severe Accident Conditions}},
697-
institution = {{Sandia National Laboratories}},
698-
address = {Albuquerque, NM},
699-
type = {{NUREG/CR-6131}},
700-
month = {December},
701-
year = {1998},
702-
note = "{Work performed under contract to the US Nuclear Regulatory Agency, Washington DC.}",
695+
author = {N.E. Bixler},
696+
title = {{VICTORIA 2.0: A Mechanistic Model for Radionuclide Behavior in a Nuclear Reactor Coolant System Under Severe Accident Conditions}},
697+
institution = {{Sandia National Laboratories}},
698+
address = {Albuquerque, NM},
699+
type = {{NUREG/CR-6131}},
700+
month = {December},
701+
year = {1998},
702+
note = "{Work performed under contract to the US Nuclear Regulatory Agency, Washington DC.}",
703703
}
704704

705705
@TECHREPORT{FAA-AR-06-21,
@@ -1877,11 +1877,11 @@ @INPROCEEDINGS{Floyd:Interflam2010
18771877
}
18781878

18791879
@misc{FPL:Fire_Database,
1880-
Author = {US Forest Service - Forest Products Laboratory},
1881-
Howpublished = {\url{https://www.fpl.fs.usda.gov/products/products/cone/by_materials.php}},
1882-
Title = {{Cone Calorimeter Datasets}},
1883-
Type = {Online Multimedia},
1884-
Year = {2008}}
1880+
Author = {US Forest Service - Forest Products Laboratory},
1881+
Howpublished = {\url{https://www.fpl.fs.usda.gov/products/products/cone/by_materials.php}},
1882+
Title = {{Cone Calorimeter Datasets}},
1883+
Type = {Online Multimedia},
1884+
Year = {2008}}
18851885

18861886
@TECHREPORT{FSSIM,
18871887
author = {J. Floyd and S. Hunt and F. Williams and P. Tatem},
@@ -2021,12 +2021,12 @@ @TECHREPORT{Friedman:1
20212021
}
20222022

20232023
@ARTICLE{Fujita,
2024-
author = {A. Fujita and R. Kurose and S. Komori},
2025-
title = {{Experimental study on effect of relative humidty on heat transfer of an evaporating water droplet in air flow}},
2026-
journal = {International Journal of Multiphase Flow},
2027-
year = {2010},
2028-
volume = {36},
2029-
pages = {244-247}
2024+
author = {A. Fujita and R. Kurose and S. Komori},
2025+
title = {{Experimental study on effect of relative humidty on heat transfer of an evaporating water droplet in air flow}},
2026+
journal = {International Journal of Multiphase Flow},
2027+
year = {2010},
2028+
volume = {36},
2029+
pages = {244-247}
20302030
}
20312031

20322032
@INPROCEEDINGS{Gandhi:2,
@@ -2058,14 +2058,14 @@ @TECHREPORT{Gavelli:1
20582058
}
20592059

20602060
@TECHREPORT{Gavin,
2061-
author = {Gavin, P.M.},
2062-
title = {{PROGRAM DROP: A Computer Program for Prediction of Evaporation from Freely Falling Multicomponent Drops}},
2063-
institution = {Sandia National Laboratories},
2064-
month = {December},
2065-
year = {1996},
2066-
type = {SAND},
2067-
number = {96-2878},
2068-
address = {Albuquerque, NM},
2061+
author = {Gavin, P.M.},
2062+
title = {{PROGRAM DROP: A Computer Program for Prediction of Evaporation from Freely Falling Multicomponent Drops}},
2063+
institution = {Sandia National Laboratories},
2064+
month = {December},
2065+
year = {1996},
2066+
type = {SAND},
2067+
number = {96-2878},
2068+
address = {Albuquerque, NM},
20692069
}
20702070

20712071
@BOOK{Gelman:Stats,
@@ -2661,11 +2661,11 @@ @ARTICLE{Harlow:1
26612661
}
26622662

26632663
@INPROCEEDINGS{Higgins_Davidson,
2664-
author = {Higgins, D. and Davidson, M.},
2665-
title = {{An Isothermal Model of Agglomeration in a Flash Smelting Reaction Shaft}},
2666-
booktitle = {Fifth International Conference on CFD in the Process Industries},
2667-
organization = {CSIRO Minerals, Melbourne, Austrailia},
2668-
year = {2006},
2664+
author = {Higgins, D. and Davidson, M.},
2665+
title = {{An Isothermal Model of Agglomeration in a Flash Smelting Reaction Shaft}},
2666+
booktitle = {Fifth International Conference on CFD in the Process Industries},
2667+
organization = {CSIRO Minerals, Melbourne, Austrailia},
2668+
year = {2006},
26692669
}
26702670

26712671
@ARTICLE{Hinkley:2,
@@ -2927,6 +2927,12 @@ @ARTICLE{Edwards:FSJ
29272927
volume = {40},
29282928
}
29292929

2930+
@MISC{HYPRE,
2931+
title = {{HYPRE: Scalable Linear Solvers and Multigrid Methods}},
2932+
note = {Lawrence Livermore National Laboratories},
2933+
howpublished = {\href{https://computing.llnl.gov/projects/hypre-scalable-linear-solvers-multigrid-methods}{HYPRE Website}},
2934+
}
2935+
29302936
@ARTICLE{Icove:JNAFE2021,
29312937
author = {D.J. Icove and T.R. May},
29322938
title = {{Computer Fire Modeling and the Law: Application to Forensic Fire Engineering Investigations}},
@@ -3302,11 +3308,11 @@ @TECHREPORT{Kim:1987
33023308
}
33033309

33043310
@INPROCEEDINGS{Kolaitis,
3305-
author = {D. Kolaitis and M. Founti},
3306-
title = {{Comparing Evaporation Rates of Single Suspended Droplets}},
3307-
booktitle = {10th Workshop on Two-Phase Flow Predictions},
3308-
year = {2002},
3309-
publisher = {European Research Community on Flow, Turbulence, and Combustion},
3311+
author = {D. Kolaitis and M. Founti},
3312+
title = {{Comparing Evaporation Rates of Single Suspended Droplets}},
3313+
booktitle = {10th Workshop on Two-Phase Flow Predictions},
3314+
year = {2002},
3315+
publisher = {European Research Community on Flow, Turbulence, and Combustion},
33103316
}
33113317

33123318

@@ -3731,12 +3737,12 @@ @ARTICLE{Madhav:IJMP1995
37313737

37323738

37333739
@ARTICLE{Maqua,
3734-
author = {C. Maqua and G. Castanet and F. Lempoine},
3735-
title = {{Bicomponent droplets evaporation: Temperature measurements and modelling}},
3736-
journal = {Fuel},
3737-
volume = {87},
3738-
pages = {2932-2942},
3739-
year = {2008}
3740+
author = {C. Maqua and G. Castanet and F. Lempoine},
3741+
title = {{Bicomponent droplets evaporation: Temperature measurements and modelling}},
3742+
journal = {Fuel},
3743+
volume = {87},
3744+
pages = {2932-2942},
3745+
year = {2008}
37403746
}
37413747

37423748
@TECHREPORT{Iowa,
@@ -5005,12 +5011,12 @@ @BOOK{Patankar:1
50055011
}
50065012

50075013
@article{Paudel:2021,
5008-
author = {D. Paudel and A. Rinta-Paavola and H.P. Mattila and S. Hostikka},
5009-
title = {{Multiphysics modelling of stone wool fire resistance}},
5010-
volume = {57},
5011-
journal = {Fire Technology},
5012-
year = {2021},
5013-
pages = {1283-1312},
5014+
author = {D. Paudel and A. Rinta-Paavola and H.P. Mattila and S. Hostikka},
5015+
title = {{Multiphysics modelling of stone wool fire resistance}},
5016+
volume = {57},
5017+
journal = {Fire Technology},
5018+
year = {2021},
5019+
pages = {1283-1312},
50145020
}
50155021

50165022
@TECHREPORT{Peacock:NBS_Multi-Room,
@@ -5596,11 +5602,12 @@ @ARTICLE{Rinta-Paavola:2023
55965602
}
55975603

55985604
@misc{RISE:Fire_Database,
5599-
Author = {RISE Research Institutes of Sweden},
5600-
Howpublished = {\url{https://firedb.extweb.sp.se/}},
5601-
Title = {{RISE Fire DataBase}},
5602-
Type = {Online Multimedia},
5603-
Year = {2005}}
5605+
Author = {RISE Research Institutes of Sweden},
5606+
Howpublished = {\url{https://firedb.extweb.sp.se/}},
5607+
Title = {{RISE Fire DataBase}},
5608+
Type = {Online Multimedia},
5609+
Year = {2005}
5610+
}
56045611

56055612
@INPROCEEDINGS{Ritchie:1,
56065613
author = {Ritchie, S.J. and Steckler, K.D. and Hamins, A. and Cleary, T.G. and Yang, J.C. and Kashiwagi, T.},
@@ -5940,11 +5947,11 @@ @ARTICLE{Smyth:1
59405947
}
59415948

59425949
@INPROCEEDINGS{Snegirev:1,
5943-
author = {A. Snegirev and V. Talavov and I. Sheinman and S. Sazhin},
5944-
title = {{An enhanced spray model for flame suppression simulations}},
5945-
booktitle = {Proceedings of the European Combustion Meeting},
5946-
year = {2011},
5947-
publisher = {Combustion Institute},
5950+
author = {A. Snegirev and V. Talavov and I. Sheinman and S. Sazhin},
5951+
title = {{An enhanced spray model for flame suppression simulations}},
5952+
booktitle = {Proceedings of the European Combustion Meeting},
5953+
year = {2011},
5954+
publisher = {Combustion Institute},
59485955
}
59495956

59505957
@BOOK{Spalding:1,

Manuals/FDS_User_Guide/FDS_User_Guide.tex

Lines changed: 3 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -9234,14 +9234,16 @@ \section{Accuracy of the Pressure Solver}
92349234
\subsection{Optional Pressure Solvers}
92359235
\label{optional_pressure_solver}
92369236

9237-
The default Poisson solver in FDS ({\ct SOLVER='FFT'} on the {\ct PRES} line) is based on the package of linear algebra routines called CrayFishpak. However, in certain circumstances you may need to use one of several alternatives that is based on the Intel\textsuperscript{\textregistered} MKL Sparse Cluster Solver.
9237+
The default Poisson solver in FDS ({\ct SOLVER='FFT'} on the {\ct PRES} line) is based on the package of linear algebra routines called CrayFishpak. However, in certain circumstances you may need to use one of several alternatives that is based on the Intel\textsuperscript{\textregistered} MKL Sparse Cluster Solver or the HYPRE solver library from Lawrence Livermore National Laboratories (LLNL) \cite{HYPRE}.
92389238
\begin{description}
92399239
\item[{\ct SOLVER='ULMAT'}] This solver is useful when the computational domain has relatively small, sealed cavities enclosed with thin (i.e. zero cell thick) walls. The small error in the normal component of velocity that you incur with the FFT solver can lead to unphysical changes in temperature, density and pressure that may eventually cause the calculation to become numerically unstable. For example, suppose you have hollow steel columns or hollow aluminum ducts with thin walls and you want to simulate the heat penetration within these spaces.
92409240

92419241
With the {\ct ULMAT} solver, the unknown values of the pressure live only in gas-phase cells, allowing for the normal components of velocity at a solid surface to be computed exactly with no penetration error. Strictly speaking, in this mode, FDS is no longer using an ``immersed boundary method'' for flow obstructions---the pressure solution is now ``unstructured,'' hence the {\ct U} in its name. This solver does not guarantee that the normal component of velocity matches perfectly at mesh boundaries, and the iterative procedure used by the default {\ct 'FFT'} solver is still used to drive the mesh interface velocity normals closer together. This solver option does allow for refinement at mesh boundaries and for internal sealed regions within the domain for which additional matrices must be computed.
92429242

92439243
The {\ct ULMAT} solver uses a substantial amount of memory per mesh. Its use should be limited to meshes of at most 150,000 cells. This number depends on the amount of random access memory (RAM) that you can devote to each MPI process, and you might want to experiment with larger meshes if you have, say, tens of gigabytes of RAM per MPI process.
92449244

9245+
\item[{\ct SOLVER='ULMAT HYPRE'}] This solver uses the same strategy as ULMAT but the solver is implemented via LLNL's HYPRE library \cite{HYPRE}. ULMAT HYPRE uses the algebraic multigrid (AMG) preconditioned conjugate gradient (PCG) solver. This solver does not require LU decomposition and storage of a dense matrix, so the memory requirements are much less than the MKL solver. Hence, ULMAT HYPRE may be considered as an option for large mesh blocks (larger than 150,000 cells). Note, however, that for small mesh blocks HYPRE is somewhat slower than MKL.
9246+
92459247
\item[{\ct SOLVER='GLMAT'}] This solver is for non-overlapping, non-stretched meshes at the same refinement level covering a single connected domain, i.e., a large rectangular box. With this solver, the pressure in both solid and gas cells is computed using an immersed boundary method for flow obstructions, the same as the default {\ct 'FFT'} solver. This mode produces the exact same pressure solution as the {\ct 'FFT'} solver would if the domain were one single mesh. That is, velocity errors at any mesh boundaries are eliminated. Note that currently a single discretization matrix is built for all gas cells defined on the model, therefore the solver is not meant to be used in cases where ther are internal sealed regions within the domain.
92469248

92479249
The {\ct GLMAT} solver is used mainly for testing and diagnostics. Because it solves the pressure over the entire computational domain, it is limited to about 1~million cells total unless your computer has an extraordinary amount of RAM.

0 commit comments

Comments
 (0)