The Development of RELAP5SCDAPSIMMOD4'0 for Reactor System Analysis and Simulation

1
The Development of RELAP5/SCDAPSIM/MOD4.0 for
Reactor System Analysis and Simulation
Presented Dr. Chris Allison Innovative Systems
Software
7th HND International Conference, Dubrovnik,  May
25 - 29, 2008
2
Outline

• Background
• General features of RELAP/SCDAPSIM/MOD4.0
• Improvements relative to earlier versions of
  RELAP5 and SCDAP/RELAP5
• Speed and reliability for wide range of
  transients
• Advantages of FORTRAN 90/95/2000 for QA and
  future development
• Improved accuracy for LWRs and other reactor
  designs
• Expanded user options
• Links to 3D reactor kinetics, integrated GUIS,
  detailed 3D TH and SA modules (Japanese SAMPSON)

3
SDTP is an international cooperative program
Background

• Nearly 60 organizations in 28 countries support
  activities
• Program members and licensed software users
• ISS administers program
• Regional technical support and training centers
  currently located in US, Europe(4), China, and
  Latin America
• Activities supported through in-kind
  contributions, membership fees, and software
  licensing fees

4
SDTP-supported software
Background

• RELAP/SCDAPSIM
• SAMPSON (NUPEC)
• Detailed severe accident analysis
• FUELSIM
• LWR fuel behavior analysis
• Advanced graphical user interfaces
• Integrated graphics displays
• ViSA (RELAP5 Visualization and Input System)
• PST (Plant simulation and training facility
  University of Mexico)
• Input analyzer

5
RELAP/SCDAPSIM/MOD4.0

• Designed to run wide range of conditions from
  normal operating up through severe accident
  transients
• Numerics and coding optimized for faster than
  real time performance
• RELAP/SCDAPSIM/MOD4 is the latest version
• Completely rewritten to FORTRAN 90/95/2000
  standards for easier model/code development and
  maintainability
• Linked to member/user supported analysis packages
• Uses publicly available RELAP/MOD3.3 and SCDAP
  models
• Extended modeling options being developed by SDTP
  members

6
Wide range of transients will run in faster than
real time on typical PC
Example of standardized PWR vessel nodalization

• Standardized multi-D plant models can be used
• Improved accuracy for more extreme TH transients
  or SAs
• Detailed SCDAP models can automatically handle
  loss of core geometry
• Simplifies input model development and QA of
  input models

7
Example - 2.5 Hour PWR station blackout transient
runs in lt10 min on typical PC
8
MOD4.0 only version of RELAP5 completely
rewritten to FORTRAN 90/95/2000 standards

• Improved portability and maintenance
• Machine dependent coding can be eliminated
• Complex data base and I/O coding can be protected
  from inadvertent changes
• Variables can be assigned more meaningful names
• Advanced mathematical/matrix operations allowed
• Model coding much easier to read and improve
• Model VV easier for QA
• Significantly reduces coding errors
• Simplifies incorporation of user supplied models
  and correlations

9
RELAP5 originally written before FORTRAN 77
standard was available
Pre MOD4

• Structured coding features not available
• if-then-else do (no indexing), do while, cycle
  and exit select case, contain..
• Resulting in
• Complex logic including large number of go to
  statements

10
Fortran pre and 77 standards did not allow
dynamic allocation of storage
Pre MOD4

• Dynamic allocation was critical to RELAP
  flexibility and performance
• Resulting in development of non-standard and
  complex coding
• Dynamic storage and reallocation of space for
  input and transient processing
• Extensive use of Equivalence to allow reasonable
  pneumonic nam

11
Dynamic storage made model development complex
and difficult to debug
Pre MOD4

• All quantities seemed to be equivalenced to the
  same few locations
• Quantities in arrays were not in consecutive
  locations but required a skip factor (stride)
• Multi-D arrays had to be mapped to a 1-D array
• Indexing and do loop indices were more
  complicated.
• Additional index variables were used for
  indexing, and separate indexes were required for
  each different offset or stride
• Ie do i iskip, jskip,kskip

12
Dynamic allocation during input process also
complex and difficult to debug
Pre MOD4

• Example Processing of hydrodynamic input
  required three blocks to be created component,
  volume, and junction blocks.
• Each block may need to be increased in size as
  each component is processed.
• Movement of other blocks is needed so that empty
  space is available for the block to expand
• As data is added to a block, the two other blocks
  may need moved to make room for the expansion of
  the block.
• Up to three moves needed for most components, but
  a complicated component such as a pump requires
  approx. 30 moves

13
MOD4 FORTRAN conversion resulted in rewriting
more than 300000 lines of coding

• Conversion required nearly 3 years to complete
• Features from Fortran-90, Fortran-95, and in one
  instance, Fortran-2000 (not yet a standard) were
  used
• Most lines of the code were modified, many
  several times, as work progressed one model at a
  time

14
Original high level structure and data
organization maintained
MOD4

• High level structure (separation of coding into
  subroutines and functions) was not changed.
• Data organization (separation of variables into
  modular blocks based on modeling feature) was not
  changed.

15
Fortran 90/95 used to eliminate machine and some
OS dependencies
MOD4

• Fortran 90 includes new intrinsic routines to
  eliminate machine/OS dependencies
• Examples - date, time, cpu usage, integer and
  floating point features and ranges

16
New Fortran features used to improve readability
and reduce coding complexity
MOD4

• Example - Derived types
• Example - Reference to liquid volume velocity in
  y direction for the ith TH volume
• PreMOD4 velf(ix1) where
• ix (i-1)ivskp filndx(4)
• MOD4 vol(i)velf(2)

17
RELAP5 thermodynamic routines were also converted
Pre-MOD4

• Early versions were written before RELAP5 was
  started, and some coding features are worse than
  those noted for RELAP5
• Non-pneumonic names were used in original coding
• S(1) temperature, s(2) pressure, s(3)
  specific volume, etc.
• a(i) specific volume at a temperature and
  pressure a(i1) internal energy a(i3)
  beta etc. Different non unit strides needed to
  increment to temperature or pressure entries

18
Thermodynamic Routines, Contd
Pre-MOD4

• Three interpolation packages in use, one each for
  light water, heavy water, and other fluids.
• Coding unique although interpolation procedures
  were supposed to be same
• Each package had to be maintained
• Two dimensional interpolation procedures were
  complex to handle effects of the saturation line

19
Interpolation routines are still complex, but
MOD4

• One interpolation package now used for all fluids
• Arguments have meaningful names st
  temperature sp pressure sv specific
  volume.
• Two dimensional arrays with meaningful names
  sttpsn(i,j)u internal energy at temperature,
  i, pressure, j.

20
MOD4.0 includes many enhanced user options to
improve accuracy or expand capabilities

• Improved modeling options
• NIST-98 water properties
• Alternative material properties including Zr-Nb
  cladding
• Fission product transport and deposition
• Detailed fuel and control rod models
• Integrated uncertainty/sensitivity analysis
• Integrated graphics including 3D and Nuclear
  Plant Analyzer displays

21
Improved Modeling Options
22
More accurate 1998 NIST formulation added as an
option

• Improved properties also improve speed and
  reliability of calculations
• Saturation line defined by thermodynamic relation
• Gibbs function for saturated liquid equals the
  Gibbs function for saturated vapor at the same
  temperature or pressure.
• Only one equation, Hemholtz function, used for
  entire single phase region

23
Preliminary Zr-Nb property option added for VVERs
and CANDUs

• Properties important for air ingression being
  incorporated by SDTP member organizations

24
Advanced fission product transport model
developed at University of Florida by Dr. Honaiser

• Vapor phenomena
• Adsorption
• Condensation
• Onto structures
• Onto aerosol surfaces
• Aerosol nucleation
• Aerosol Phenomena
• Deposition
• Agglomeration
• Re-suspension

25
Improved fuel and control rod behavior models
being developed by SDTP members based ongoing EU
research programs
Quench Experiments ( FzK)

• Impact of oxidation state and fuel rod
  temperature on hydrogen source term during
  quenching
• Impact of steam starvation on the bundle behavior
  during cooldown phase
• Impact of control rods on bundle degradation and
  gas generation
• Database for validation

26
Improved fuel and control rod behavior models
being developed by SDTP members based ongoing EU
research programs
Irradiated Fuel
Fresh Fuel
Ag-In-Cd Control rods
Phebus Experiments
Zircaloy Stifners

• Behavior of irradiated fuel bundles during
  heating and melting transients
• Fission product release and transport

Phebus Fuel Bundle
27
Improved horizontal CANDU fuel channel models
being developed at the Poletechnic University in
Bucharest
28
Integrated uncertainty/sensitivity analysis
packages being developed at University of
Catalunya in Barcelona and University of Pisa
29
Graphical Packages Interfaced with MOD4
30
3D Display of TRIGA input model
         
31
Display of Lower Plenum Model for Surry PWR input
model
32
VISA Nuclear Plant Analyzer Display
33
PST provide simulation GUI for multiple desktop
student and instructor training environment
34
MOD4.0 much easier to couple to other analysis
packages

• User supplied modules
• 3D reactor kinetics
• BWR containment analysis
• RELSIM nuclear plant analyzer display
• SDTP supplied modules
• Nuclear plant analyzer GUIS including VISA
  (KAERI) and PST (University of Mexico)
• Subchannel TH analysis
• SAMPSON (NUPEC-Japan) SA analysis modules
• Steam explosion
• Melt spreading
• H2 mixing