2.2. CSASShift: Criticality Safety Analysis Sequence with Shift
K. B. Bekar, G. Davidson, B. Langley, B.J. Marshall
The CSASShift sequence integrates the Shift advanced Monte Carlo solver into the CSAS framework as an alternative to the KENO transport solvers to perform reliable and efficient eigenvalue calculations for criticality safety and reactor physics analysis. It supports both KENO V.a and KENOVI geometries and provides most widely used KENO capabilities available in the CSAS5 and CSAS6 sequences for both multigroup and continuousenergy transport modes. The highly scalable Shift Monte Carlo solver enables faster solutions when running on multiple cores, and it shows better performance to achieve the same level of accuracy compared to the CSAS sequences with the KENO codes.
2.2.1. Introduction
The CSASShift sequence was designed to provide the modeling and simulation capabilities required for criticality safety and reactor physics analysis through the Shift Monte Carlo solver.
Shift is a massively parallel Monte Carlo radiation transport package in the Exnihilo radiation transport code suite [CSASShiftPJE+16], [CSASShiftESSC10]. Shift was developed including features such as support for both fixedsource and eigenvalue Monte Carlo transport capabilities with multiple geometry and physics engines, hybrid capabilities for variance reduction methods, and advanced parallel decompositions to scale well from laptops to small computing clusters to advanced supercomputers. Shift supports different geometry engines, including the Oak Ridge Adaptable Nested Geometry Engine (ORANGE) designed to provide particle transport capabilities on both KENO V.a and KENOVI geometries as well as geometry visualization capabilities in the Fulcrum user interface. Shift with both versions of KENO geometries performs eigenvalue calculations in both continuousenergy and multigroup modes. Shift supports most widely used primary capabilities available with the KENO codes and provides some unique capabilities with ORANGE such as modeling randomly packed media and efficient parallel calculations for volume estimates.
CSASShift provides all capabilities for both multigroup and continuousenergy transport modes. Like the CSAS5 and CSAS6 implementation, in the multigroup calculation mode, CSASShift sequences automate the processing of the cross sections for temperature corrections and problemdependent resonance selfshielding for utilization in multigroup neutron transport calculations using SCALE’s cross section processing module, XSProc. If continuousenergy calculation mode is selected, no resonance processing is needed, and the continuousenergy cross sections are used directly in the Shift code, with temperature corrections provided as the cross sections are loaded.
CSASShift with the highly scalable Shift solver enables some unique capabilities and faster solutions when running on multiple cores, and it shows better performance to achieve the same level of accuracy compared to the CSAS sequences with the KENO codes. CSASShift input requirements, supported and unsupported capabilities, and input and output details are described in the following sections.
2.2.2. CSASShift Input Requirements
CSASShift’s design is aimed to make a smooth transition between
KENO codes to the Shift transport code. Therefore, the original input
data layout available in CSAS sequence with the KENO transport codes
was kept the same for the CSAS sequence with Shift transport.
CSASShift uses the same CSAS5 and CSAS6 inputs, the only input
modification that should be required is changing the sequence
name by appending shift
to the sequence name, as shown in
Example 2.2.1.
Like CSAS sequences with KENO codes, CSASShift sequences are named with the KENO geometry that they support: CSAS5Shift for the models with KENO V.a geometry, and CSAS6Shift for the models with KENOVI geometry.
=csas5
Godiva sphere
ce_v7.1
read composition
u234 1 0 0.000491995 300 end
u235 1 0 0.0449996 300 end
u238 1 0 0.002498 300 end
end composition
read parameter
html=no
end parameter
read geometry
sphere 1 1 8.741
end geometry
end data
end
=csas5shift
Godiva sphere
ce_v7.1
read composition
u234 1 0 0.000491995 300 end
u235 1 0 0.0449996 300 end
u238 1 0 0.002498 300 end
end composition
read parameter
html=no
end parameter
read geometry
sphere 1 1 8.741
end geometry
end data
end
Warning
CSASShift sequence implementation may ask the user for some minor input updates for a successful calculation.
2.2.3. Sequence Capabilities
In the CSASShift sequence framework, SCALE data handling is automated as much as possible. Similar to many other SCALE sequences, CSASShift also applies a standardized procedure to provide appropriate number densities and cross sections for the calculation. XSProc is responsible for reading the standard composition data and other engineeringtype specifications—including volume fraction or percent theoretical density, temperature, and isotopic distribution, as well as the unit cell data. XSProc then generates number densities and related information, prepares geometry data for resonance selfshielding and fluxweighting cell calculations, and (if needed) provides problemdependent multigroup cross section processing.
Sequences that execute Shift transport include a data processor named ExnihiloInputBuilder to read and check the KENO data. This data processor processes the KENO data and creates a ParameterList input used by Shift to construct the problem and perform transport calculations. When the data checking has been completed, the CSASShift sequence executes XSProc to prepare a resonancecorrected macroscopic cross section library in the AMPX working library format for the subsequent Shift transport calculation if a multigroup library has been selected.
Similar to CSAS sequences with KENO transport, the CSASShift sequence supports both CELLMIX and Doublehet capabilities. For each unit cell specified as being cellweighted, XSProc performs the necessary calculations and produces a cellweighted macroscopic crosssection library. Shift may be executed to calculate the keffective, or neutron multiplication factor, using the cross section library that was prepared by the control sequence.
Computational capabilities available in CSAS sequences with KENO codes—including the determination of keffective, flux densities, fission densities, mesh tallies, Shannon entropy tally, problemdependent continuousenergy temperature treatments, parallel calculations, and many more—are also provided by the CSASShift sequence.
CSASShift also supports two new CSAS sequence data blocks, definitions and tallies data, to allow flexible definition and output control of mesh tallies. The mesh responses neutron flux, fission rate, and fission source can now be requested multiple times on different spatial and energy grids in the same calculation. This capability helps users efficiently manage computational resources when collecting detailed information, depending on their requirements.
Criticality safety tools in SCALE attain some unique capabilities provided by Shift with the new geometry engine ORANGE, such as parallel volume estimates for KENOVI geometric regions and modeling randomly packed media, which is enabled by implementing a randompacking algorithm to place spherical particles within simple bounding geometries. This capability allows constructing tristructural isotropic (TRISO) particle models for advanced reactor modeling and simulation activities. See Sect. 2.2.4.1.2.1 for further details.
Details for some of these capabilities, their input methods, and output edits are provided in the following sections.
2.2.3.1. Multigroup limitations
Some of the limitations of the CSASShift multigroup sequences are a result of using preprocessed multigroup cross sections. Inherent limitations in multigroup CSASShift calculations are as follows:
Spatial effects such as fuel rods in assemblies where some positions are filled with control rod guide tubes, burnable poison rods, and/or fuel rods of different enrichments. The cross sections are processed as if the rods are in an infinite lattice of identical rods. If the user inputs a Dancoff factor for the cell (such as one computed by MCDancoff), XSProc can produce an infinite lattice cell which reproduces that Dancoff. This can mitigate some spatial lattice effects.
2.2.3.2. Continuousenergy limitations
The continuousenergy cross sections are directly used in Shift. An existing multigroup input file can easily be converted to a continuousenergy input file by simply specifying the continuousenergy library. In this case, all cell data is ignored. However, the following limitations exist:
If CELLMIX is defined in the cell data, the problem will not run in continuousenergy mode. CELLMIX implies new mixture cross sections are generated using XSDRNPMcalculated cell fluxes; therefore it is not applicable in continuousenergy mode.
Problems with DOUBLEHET cell data are not allowed, as they inherently utilize the CELLMIX feature.
2.2.3.3. Unsupported Capabilities
Although Shift Monte Carlo code was designed with several advanced capabilities, it does not currently support some of the unique features available in KENO codes. Therefore, CSASShift does not provide some of the capabilities available in CSAS5 and CSAS6 sequences.
The missing capabilities are mostly considered as the outdated features or those seldomly used by CSAS users in their analysis. The equivalent capabilities will be activated in the Shift transport code in the next SCALE release, depending on the need basis.
Table 2.2.1 summarizes the capabilities currently supported by CSAS with KENO codes but not supported by CSASShift sequences.
Capability 
Input method(s) to activate the capability 
Comments 

Adjoint transport 
parameter data 
Adjoint transport capability is not available in Shift 
Promptonly \(nu\) 
parameter 
Shift does not support using prompt neutron spectrum only in continuousenergy mode 
Use unionized mixture cross section 
parameter 
Shift does not support KENOlike mechanism to store cross sections on a materialbased unionized energy grid for a faster cross section lookup in continuousenergy mode Although this method benefits for faster runtimes for some KENO models, storing all data may require prohibitively large amount of memory for problem with a large number of materials. Different approaches are being developed in Shift transport, and some experimental implementation is available in CSASShift. See Table 2.2.4 for more details. 
Matrix keffective 
parameters

An alternative keffective calculation method available in KENO codes are not supported by Shift. 
Start data types 2, 3, 4, 5 and 9 

Start data types 2  5 have not been implemented by ORANGE geometry engine used by Shift transport. Start type 9 designed to read starting distribution from a mesh source file is not currently supported by CSASShift. 
Biasing or weighting data 

KENOlike biasing capability is not currently supported by Shift transport. 
Periodic and White Albedo boundary conditions 

Shift transport does not currently support PERIODIC and WHITE boundary conditions for both KENO V.a and KENOVI geometries. 
Differential Albedos 

Material specific albedos available with KENO multigroup transport is not supported by Shift multigroup transport 
LOOP construct in array data 

LOOP construct in array data input block is not supported by ORANGE geometry engine as part of Shift code. 
Volume calculation (random sampling) 

Random volume estimates for KENOVI geometry is not available in ORANGE geometry engine used by Shift transport 
Accumulate mesh fluxes 
parameter 
Shift does not support to tally mesh fluxes which are averaged over the region volumes in each mesh voxel. 
Compute and print mean free paths 
parameter 
This capability is not currently implemented in Shift. 
Regiondependent fissions and absorptions 
parameters 
Although these tallies are available in Shift transport they are not currently implemented in CSASShift. 
Mixturedependent reaction tallies 

Although these tallies are available in Shift transport they are not currently implemented in CSASShift. 
Time controlled termination 
parameter 
Shift does not have job termination capability controlled by the userdefined time limit. 
Terminate execution on user signal 
by creating a file named

CSASShift does not support this capability 
Restart capability 
parameters 
Restart capability is not available in Shift 
Print particle tracks 
parameter 
Although Shift has its own mechanisms to print information about each particle history, this capability is not fully integrated in CSASShift sequence. It will be available in next releases. 
Problem Characterization Output Edit 
always ON 
This capability is not currently implemented in CSASShift. 
Frequency Distributions 
always ON 
This capability is not currently implemented in CSASShift 
Plots of avg. keffective and Shannon entropy 
always ON 
CSASShift does not plot the average keffective by generations run, the keffective by generations skipped, and Shannon entropy per generation. Instead of the charplots, CSASShift creates Ptolemy plots and stores them in a dedicated plot directory. Fulcrum may be used to visualize these plots. See Sect. 2.2.6.1.12 
Summary of Source Convergence Diagnostics 
parameter 
CSASShift does not perform the posterior entropy tests available in KENO. Instead, result of a single test performed by Shift is captured and printed in the relevant output section. 
Print capabilities for mixed cross section 
parameters 
Currently, no capability is available to print the cross sections used by Shift transport. 
Flux moments and angular flux calc. 
parameters 
CSASShift does not support any of these capabilities. 
Print starting points 

This capability is not currently implemented in CSASShift 
Plot capability 
parameter 
Oldstyle plotting capabilities available in KENO codes are not supported. Fulcrum can be used for geometry visualization. 
HTML output 
parameter 
Oldstyle HTMLbased output method is not supported. 
2.2.4. Input Data Guide
This section describes the input data required for the CSASShift
sequence. A typical CSAS6Shift input, shown in
Example 2.2.2, starts
with the sequence identifier always preceded by the =
sign, and
it is followed by the problem title. Then, a cross section library
name is specified, and all these entries are followed by several
data blocks each starting with READ data_block and
ending with END data_block.
=CSAS6Shift parm=(parm_options)
problem title
'  XSProc data
' cross section library name (REQUIRED)
ce_v7.1
' List of material specifications in standard SCALE format (REQUIRED)
read composition
...
end composition
' Specify data for resonance processing (OPTIONAL)
read celldata
...
end celldata
'  New CSAS sequence data blocks
' Used to define energy bounds and grid geometries for
' the tallies defined in tallies data block
' (REQUIRED if tallies data block exists)
read definitions
...
end definitions
' Used to define tallies in a more robust way (OPTIONAL)
read tallies
...
end tallies
'  KENO transport data
' Specify the problem geometry (REQUIRED)
read geometry
...
end geometry
' Other input data blocks (OPTIONAL)
Because CSASShift uses the same input data used by CSAS5 and CSAS6, details of the input data blocks, compositions, celldata, definitions, and tallies will not be repeated here, and they can be seen in Sect. 2.1.4. Data blocks in the KENO transport data category will be discussed in the following section.
Note
For CSASShift, the grid boundaries must be inside the specified geometry, while CSASKENO permits grid boundaries beyond the geometry.
2.2.4.1. KENO input data in CSASShift
Table 2.2.2
presents the lists of the KENO input data blocks supported by CSASShift
sequence. The input method in some data blocks may show some minor differences
between the CSASKENO and CSASShift sequences. Similarly, some capabilities
provided by each input block also have some differences. All these details
are discussed in this section. KENO input data blocks, that are
reaction data
,
bias data
, mixt data
, and plot data
, are not currently
supported by the CSASShift sequences.
Data block 
Status 
Comments 
Parameters^{*} 
Supported 
See Sect. 2.2.4.1.1 for more details. 
Geometry 
Supported 
See Sect. 2.2.4.1.2 for more details. 
Array data 
Supported 

Boundary conditions 
Supported 
See Sect. 2.2.4.1.2 for more details. 
Volume data 
Supported 
See Sect. 2.2.4.1.2 for more details. 
Energy boundaries 
Supported 

Start data 
Supported 
See Sect. 2.2.4.1.3 for more details. 
Grid geometry data 
Supported 

Plot data 
Not available 
The data in this block will be ignored in the calculations 
Bias data 
Not available 
Execution will be terminated. 
Reaction data 
Not available 
The data in this block will be ignored in the calculations 
Mixt data 
Not available 
The data in this block will be ignored in the calculations 
^{*} Must precede all other data blocks in this table. 
Similar to CSAS5 and CSAS6, geometry data
is the only KENO
data block required to perform Shift transport calculation as part of the CSASShift
sequence. Other data blocks are optional, and the same default values
listed in various locations in Sect. 8.1 are also applied to the
data in each data block in CSASShift. Note that parameter data must
precede all other KENO data blocks if it is entered.
When CSASShift is run with a user input including
bias data
block, the execution will be terminated with the error
message given in Example 2.2.3.
Note that CSASShift ignores the data entered in the unsupported
plot data
, mixt data
, and reaction data
blocks and
continues the calculation. User is notified with a warning message as
shown in Example 2.2.4.
***Error: Failed to run ExnihiloModule with assertion:

These input cards are unsupported by Shift.

They must be removed from the input to run.

line: 19 column: 1 biasing

^^^ at /ornldev/code/Scale/S63/Source/packages/Module/Exnihilo/InputProcessorBase.cpp:193
====================================================================================================
Input Warnings:
====================================================================================================
***Warning: Plot block found. This is not currently supported by Shift and will be ignored for now.
Note
CSAS5Shift and CSAS6Shift also support PARM=CHECK or PARM=CHK sequence parm options. This will allow checking the input data without performing cross section calculation as well as Shift transport calculations.
=CSAS6SHIFT PARM=CHK
2.2.4.1.1. Parameter data
The KENO parameter data block in both CSAS5 and CSAS6 sequences provides many control parameters to activate the capabilities available in KENO transport for the problem being run. CSASShift supports only a subset of these parameters, as listed in Table 2.2.3. Detailed description of these parameter can be seen in Sect. 8.1.3.3.
Parameters entered in the parameter data
input block are processed by
CSASShift sequence implementation, and then the ParameterList input
is updated to accordingly activate/deactivate the equivalent capabilities
with Shift transport if the asking feature is currently supported by Shift.
CSASShift usually ignores the unsupported parameters by notifying the user
with a warning message, and then it continues the calculation. For some
specific parameters, code can terminate the execution and ask the user to
remove this parameter from the input and rerun the code for a successful
calculation.
Caution
CSASShift notifies users of the unsupported parameters with a warning message before Numeric and Logical Parameters edit in the output, and then it ignores this parameter. It is the user’s responsibility to examine which input parameter is ignored in the current calculation.
Note
CSASShift defaults the value of a parameter, which is currently
supported but not defined in the parameter data
input block, to the KENO
default. In other words, both CSAS and CSASShift use the same defaults
for the same parameters.
PARAMETERS: 
Format: See Sect. 8.1.3.3 for details. 


KEY 
DEFAULT 
DEFINITION 
KEY 
DEFAULT 
DEFINITION 

given 
random number 

10.0 
thermal energy cutoff (eV) 

0.0 
deviation limit 

0.4 
DBRC lower energy cutoff (eV) 

0.5 
average weight 

210 
DBRC upper energy cutoff (eV) 

1/WTH 
Russian Roulette weight 

0.0 
mesh size of the cubic mesh 

203 
number of generations 

0 
use DBRC for scattering 

1000 
number per generation 

2 
Doppler Broadening method 

3 
generations skipped 

0 
CE TSUNAMI calculation mode 

252 
number of energy groups for tallying 

1 
number of latent generations for CE SUNAMI 

YES 
use probability tables 

YES 
fission source convergence diag. 

NO 
collect and print region fluxes 

NO 
accumulate neutron production 

YES 
fission densities 

NO 
fission rate mesh tally 

YES 
neutrons per fission 

NO 
compute grid fluxes 

NO 
print F*(r) 3dmap 

NO 
use mesh for CLUTCH F*(r) calc. 

YES 
execute problem 

NO 
NOT FULLY IMPLEMENTED 
PARAMETERS: 


KEY 
DEFAULT 
DEFINITION 

5 
Legendre polynomial order 

0.0 
Accelerate xsec calculation using double indexing 

0.0001 
Tolerance to use thinning the unionized xsec grid 
In CSAS5 and CSAS6, users can control the number of scattering angles
in multigroup calculations by entering the SCT
parameter in the KENO
mixing data
block. The similar capability in CSASShift was
provided by adding a new parameter, PN_ORDER=
, to the parameter data
block because the mixing data
block is not supported by CSASShift sequences.
Its default is set as 5.
2.2.4.1.2. Geometry data
CSASShift with ORANGE geometry supports all KENO V.a and KENOVI geometry capabilities except the following:
LOOP
construct inarray data
blockMaterialspecific albedo boundary conditions
PERIODIC and WHITE albedo boundary conditions
Random volume estimates for KENOVI geometry (
TYPE=random
option involume data
)
CSASShift can perform volume calculations with the stochastic raytracing method concurrently on the replicated domain on multiple cores.
2.2.4.1.2.1. Random geometry
Another new capability in CSASShift is the automatic placement of units within another unit. This capability is currently limited to the stochastic placement of spherical geometries without clipping within another geometry. Additional options such as the automatic placement in lattice structures and the extension to other geometries is planned for future SCALE releases.
This new capability is enabled through a new input block named randomgeom
.
The randomgeom
block is composed of randommix
specifications which is
again composed of key/value pairs. The basic structure of the randomgeom
block and the randommix
specifications is as follows:
read randomgeom
RANDOMMIX = ID
TYPE=random
UNITS= U1 U2 ... UN end
PFS= pf1 pf2 ... pfN end
CLIP= no
SEED= int
end RANDOMMIX
end randomgeom
with
RANDOMMIX  keyword with ID number or name
TYPE  distribution type of units (currently limited to random)
UNITS  list of unit number(s) to be distributed in geometry
PFS  fraction of volume occupied by units
U1 ... UN
CLIP  boundary clipping (currently limited to no)
SEED  random seed for random placement
Similar to the array block, this input block requires that the unit
specified as part of a randommix in the units list must exist in the geometry.
An additional requirement is that the the units must have a spherical outer boundary.
The PF
list must have the same length as the units list. The sum of the values
listed in PFS
must be less than 1.0. In practice, the actual limit to the
total PF depends on the size and number of the units specified by the user.
The TYPE
keyword specifies the distribution of the units within the geometry.
The type is currently limited to random which will call a stochastic
placement algorithm to randomly distribute the units in the geometry.
The CLIP
keyword controls the clipping of the units along the geometry in
which they are placed. This is currently limited to no, that is the units
are not clipped by the geometry. The SEED
keyword is specifying the random
seed used for the stochastic distribution of units. This assures the same random
distribution if the input is run multiple times.
Note
Given the stochastic algorithm that is currently called by the randommix block, in practice total packing fractions of up to approximately 20% are achieved.
A “fill” material—the interstitial media surrounding the random spherical
geometry units—is not present in the randomgeom
block. Instead, the fill
media is handled in the unit specification within the KENO geometry
block of the input file: A region in a unit is filled with both a media and a
randommix
record. Then the media is filling the space of the region that is
not occupied by units placed through the units defined in the randommix
record.
A randommix
block can be used in multiple different units with varying fill
materials. The basic unit format along with a randommix
on the media record itself is as follows:
Note
A randommix
can currently be filled only into regions that have
an outer boundary of a sphere, cuboid, or cylinder.
read randomgeom
randommix=ID
...
end randommix
end randomgeom
read geometry
...
unit U
surfaces ...
media ...
media F biasID surfaces randommix=ID
boundary S
...
end geometry
In the sample case shown in Example 2.2.6, a single pebble is filled with a single TRISO particle type. Fig. 2.2.1 shows how TRISO particles are placed in a single pebble with the randomgeom capability. The individual location of particles is written in the Shift Hierarchical Data Format (HDF5) output file, so these locations can be used for verification or other purposes as needed.
=csas6shift
...
read geometry
unit 1
com='kernel1'
sphere 1 2.50e02
sphere 2 3.40e02
sphere 3 3.80e02
sphere 4 4.15e02
sphere 5 4.55e02
media 100 1 1
media 101 1 2 1
media 102 1 3 2
media 103 1 4 3
media 104 1 5 4
boundary 5
global unit 10
com='pebble'
sphere 1 2.5
sphere 2 3.0
cuboid 3 6p5.0
media 101 1 1 RANDOMMIX='trisos'
media 106 1 2 1
media 0 1 3 2
boundary 3
end geometry
read randomgeom
randommix = 'trisos'
type= random
units= 1 end
pfs= 0.05 end
clip= no
seed= 0
end randommix
end randomgeom
end data
end
2.2.4.1.3. Start data
CSASShift supports only START types 0, 1, 6, 7, and 8.
CSASShift start data implementation does not currently
support the PSP
option, which is used to print
source positions sampled by the Shift transport.
Implementations for start types 0, 1, 7, and 8 in CSASShift are are similar to those in CSAS5 and CSAS6. However, there are some minor differences in start type 6.
In KENO start type 6 implementation, the following rules are applied when selecting the starting points (see Sect. 8.1.4.8 and Sect. 8.1.3.3 for more details).
Start NPG initial fission neutrons at firstNPG starting points defined by start type 6 data if NPG < LNU. Remaining starting points beyond NPG will be discarded.
Start NPG initial fission neutrons at LNU starting points defined by start type 6 data if NPG = LNU.
Start LNU initial fission neutrons at the starting points defined by start type 6 data, then randomly select the remaining fission source points (NPGLNU) from these starting points if NPG > LNU.
where LNU, a start type 6 data parameter, is the total number of starting points specified in the start data block; and NPG, a parameter in the parameter data block, is the number of neutrons per generations.
Unlike KENO, the CSASShift input processor does not follow the above rules when selecting positions for the initial fission neutrons. It calculates the probability of each point being selected and passes all starting points with this information to the Shift module. The Shift module always samples NPG initial fission source points using these data.
For example, the KENO code processes the following input and then samples the initial fission points.
=csas6 parm=bonami
Godiva test problem
test8grp
read composition
u234 1 0 0.000491995 300 end
u235 1 0 0.0449996 300 end
u238 1 0 0.002498 300 end
end composition
read parameter
htm=no gen=10 npg=15
end parameter
read geometry
global unit 1
sphere 1 8.67
media 1 1 1
boundary 1
end geometry
read start
nst=6 ps6=yes psp=yes
tfx=1.0 tfy=1.0 tfz=1.0 lnu=5
tfx=2.0 tfy=2.0 tfz=2.0 lnu=20
end start
end data
end
The summary of the sampling process is printed in KENO output, as shown in Fig. 2.2.2. KENO first starts 5 neutrons at (1.0, 1.0, 1.0) and the remaining 10 neutrons at (2.0, 2.0, 2.0) since LNU=20 > NPG=15.
However, CSASShift creates a probability distribution from the defined start type 6 points and samples starting positions for NPG=15 particles using this distribution, as shown in Fig. 2.2.3.
2.2.5. Notes to CSASShift Users
In Monte Carlo calculation, the variance of the eigenvalue (keffective) at each generation is calculated as a sample variance, which is the quantity obtained by assuming no correlation over the generations. However, there is a correlation among the fission sources over generations since the deviation of fission source at a generation from its equilibrium distribution is transferred to the following generations. To resolve this issue, KENO codes use an iterative method to estimate the real variance. [CritSafetyUMN97]
The same methodology is also implemented inside Shift transport, so that printed keffective uncertainties in both message and output files are derived from the real variance estimates.
Caution
User may observe differences in keffective uncertainty values estimated by CSAS and CSASShift sequences when running the identical problem. This is mainly due to the use of a different keffective estimator in both KENO and Shift transport codes. Note that, KENO implements absorption estimator in CE mode and collision estimator in MG mode, whereas Shift implements tracklength estimator for both CE and MG modes.
Warning
keffective values and associated information from Shift
calculation and some diagnostics messages originated by Shift are
always printed to the standard output (and .msg
file).
There is no user option to suppress these.
2.2.6. CSASShift Output
The CSASShift sequence output is similar to the CSAS5 and CSAS6 outputs, except the output section dedicated to the transport module. See Sect. 2.1.5 for the layout of the output, mixture table edit, and cross section processing summary edits for a typical CSAS sequence.
This section contains a brief description of the output section dedicated to the Shift transport module. This section provides representative samples of the output format. The actual data contained in this section are not necessarily consistent with results computed by the current version of CSASShift.
2.2.6.1. Shift output
The output layout of the Shift transport module is generally similar to the typical KENO output edits printed in the CSAS sequence output. However, some output edits are printed in a very different format, and these are discussed in this section.
Warning and error messages show stylistic differences compared to the traditional CSAS and KENO messages, and their details are not documented in this manual.
2.2.6.1.1. Program verification information
Program verification information Fig. 2.2.4 is printed after the header page. It lists the name of the program, the date the load module was created, the library that contains the load module, the computer code name from the configuration control table, and the revision number. The job name, date, and time of execution are also printed. This information may be used for quality assurance purposes.
2.2.6.1.2. General problem information
A general problem information output edit, shown in Fig. 2.2.5, follows the program verification information table. This table is printed by all SCALE Shift sequence implementations. After printing the title given in the input, it summarizes some highlevel information for the physics setup of the Shift code.
2.2.6.1.3. Input Warnings
CSASShift captures the warning messages emitted from the ExnihiloInputBuilder when processing KENO data for the Shift transport. All these stacked warning messages are printed in the input warnings output table as shown in Fig. 2.2.6.
2.2.6.1.4. Tables of parameter data
The CSASShift parameter edits list both numeric and logical parameters in the same table. In each table row, the name of the KENO parameter, its short description, the current value of the parameter, and its input method are printed. If the parameter value has been entered by user in the KENO parameter data block, the input method is printed as ( input * ). Otherwise, the input method is printed as ( default ). The user should always verify that the parameter data block was entered as desired. An example of the parameters table is shown in Fig. 2.2.7.
2.2.6.1.5. Energy boundaries data
The CSASShift implementation supports multiple sets of energy boundaries specifications for some of the tallies. This can be done by using the definitions data block as described in the CSAS manual Sect. 2.1.4. However, it prints only the default energy group bounds in the energy boundaries data edit, as illustrated in Fig. 2.2.8. Energy group boundaries used for each mesh tally will be printed in the mesh tallies output edit.
2.2.6.1.6. Mixing table data edits
CSASShift uses the same format and contents as those described for KENO codes in Sect. 8.1.3.10 for the mixing table data edits. In this table, the mixture number, density, and temperature are first printed, followed by a table of the nuclides which make up the mixture. This table contains the following data: nuclide ID number, nuclide mixture ID number, atom density, weight fraction of nuclide in mixture, ZA number, atomic weight, temperature, and nuclide title. Mixture temperature is the same as the nuclides’ temperatures for the multigroup calculations, but it may show some differences in continuousenergy calculations. See Sect. 8.1.3.10 for details.
A sample mixing table data edit is shown in Fig. 2.2.9 for a multigroup calculation.
2.2.6.1.7. Geometry data edits
CSASShift captures the output edit from ORANGE and prints these data as the overview of the geometry. Its format is completely different from the traditional KENO geometry output format but includes more descriptive sections for each geometry piece, as shown in Fig. 2.2.10.
2.2.6.1.8. Volume information
Volume tables for both KENO V.a and KENOVI geometries are always printed by CSASShift using the KENOstyle volume editing format and cannot be suppressed. KENO V.a and KENOVI volume tables show some differences, and all these details are described in KENO manual. See Sect. 8.1.5.17 for further details.
A sample volume output edit for KENOVI geometry printed by CSASShift is shown in Fig. 2.2.11.
2.2.6.1.9. Initial source edits
A summary table is always printed for start types 0, 1, 6, 7, and 8.
The table format is the same for both KENO V.a and KENOVI geometries.
Fig. 2.2.12 illustrates typical starting
data for start type 0. The parameter used in this example was NST=0
.
2.2.6.1.10. Keffectives by generation
At the completion of each generation, CSASShift prints the keffective for that generation and associated information obtained from the Shift transport module. An example of this printout is given in Fig. 2.2.13.
The data printed include (1) the generation number, (2) the keffective
calculated for the generation, (3) the average value of keffective
through the current generation (excluding the nskip
1 generations),
(4) the deviation associated with the average keffective, and (5) Shannon
entropy for the generation.
After the last generation, a message is printed to indicate why execution was terminated. The user should examine this portion of the printed results to ensure that keffective is in acceptable agreement and to verify that the average value of keffective has become relatively stable. If the keffectives appear to be oscillating or drifting significantly, then the user should consider rerunning the problem with a larger number of histories per generation.
Note
keffective values from Shift calculations are always printed
to the standard output (and .msg
file). There is no user
option to suppress this.
2.2.6.1.11. Final keffective edit
The final keffective edit prints the average keffective, its associated deviation, and the limits of keffective for the 67, 95, and 99% confidence intervals. The number of histories used in calculating the average keffective is also printed. This is done by skipping various numbers of generations. The user should carefully examine the final keffective edit to determine whether the average keffective is relatively stable. If a noticeable drift is apparent as the number of initial generations skipped increases, then it may indicate a problem in converging the source. If this appears to be the case, the problem should be rerun with a better initial source distribution and should be run for sufficient number of generations so that the average keffective becomes stable. The final keffective edit is printed as shown in Fig. 2.2.14.
2.2.6.1.12. Plot of average keffective by generations run and by generations skipped
ASCII character plots of the average keffective versus
the number of generations run, and the average keffective versus
the number of generations skipped, are not printed by CSASShift
in the code output. Instead, two Ptolemy plot files are created
and copied into the plots directory in ${OUTDIR}
. The name of
the plot files and their final destinations are printed in
the output, followed by the final keffective edit as illustrated in
Fig. 2.2.15. These plot files can be
loaded and visualized by Fulcrum.
2.2.6.1.13. Shannon Entropy Diagnostics
CSASShift does not perform any posterior entropy tests like those available in KENO codes. Instead, it captures diagnostic test results performed by Shift and prints its details in the Shannon entropy diagnostics output edit, as shown in Fig. 2.2.16
2.2.6.1.14. Fission densities
The fission density edit is optional. CSASShift prints
the neutron production density and the fission density
for each geometry region if parameters FDN=YES
and NUB=YES
are specified in the parameter data (these are the default values).
If NUB=NO
is specified but FDN=YES
, then only the
production density will be given. An example of the fission
density edit is shown in Fig. 2.2.17
2.2.6.1.15. Flux Edit
Printing the fluxes is optional; they are printed only if
FLX=YES
is specified in the parameter data.
The fluxes are printed for each unit and each
geometry region in the unit for every energy group.
A sample of a flux edit is given in Fig. 2.2.18.
2.2.6.1.16. Final results table
The final results table contains the bestestimate system keffective with one standard deviation, the number of warning and error messages generated during code execution, and a final statement on the convergence of the \(\chi^2\) test results as shown in Fig. 2.2.19.
2.2.6.1.17. Final timing report table
The final timing report table summarizes the time elapsed for input processing, cell processing (for multigroup mode), cross section processing (for multigroup mode), the entire transport process (Shift transport), and postprocessing performed by the CSASShift sequence after obtaining all results from the Shift transport calculation. A sample timing report obtained for a multigroup calculation is shown in Fig. 2.2.20.
References
 CSASShiftESSC10
Thomas M. Evans, Alissa S. Stafford, Rachel N. Slaybaugh, and Kevin T. Clarno. Denovo: A new threedimensional parallel discrete ordinates code in SCALE. Nuclear technology, 171(2):171–200, 2010.
 CSASShiftPJE+16
T. M. Pandya, S. R. Johnson, Evans, T.M., G. G. Davidson, S. P. Hamilton, and A. T. Godfrey. Implementation, capabilities, and benchmarking of shift, a massively parallel monte carlo radiation transport code. Journal of Computational Physics, 308:239–272, 2016. Publisher: Elsevier.