General and Advanced Simulation
General and Advanced Simulation
VERSION 2016
by
Gamma Technologies
GT SUPPORT
TELEPHONE: (630) 325-5848
E-MAIL: support@gtisoft.com
TABLE OF CONTENTS
1.1.1 Applications
GT-SUITE is a single software package for modeling and simulation of systems in automotive and
transportation engineering and beyond. It is based on a multi-physics platform, but offers higher-level,
added-value toolboxes specifically tailored to a continually broadening set of applications:
GT-SUITE, being a single package, allows modeling of these applications both independently, as has
traditionally been the norm in the industry, or with several of these systems in one model to get the full
integrated effect of the system. For example, consider the cooling circuit with a charge-air-cooler &
EGR cooler and its link to engine performance; this can be readily analyzed in a single model for warm-
up analysis & for the design of a combined cooling circuit and engine.
Even though GT-SUITE is a single software program, additional licensing options are available at lower
costs for organizations that do not require the full capabilities of GT-SUITE by selectively deactivating
certain templates. The following licensing options are available:
GT-SUITE - full software suite. (The majority of OEM licensees choose this option.)
GT-POWER - focuses on engine performance, gas acoustics, aftertreatment (template libraries
for hydraulics, cranktrain dynamics, valvetrain dynamics, and advanced cooling are disabled)
GT-SUITEmp - full GT-SUITE minus detailed cylinder modeling and combustion analysis
GT-POWERLab - focuses on in-cylinder analysis only. It is a subset of GT-POWER, with
turbo analysis, predictive combustion, etc. disabled.
Gamma Technologies © 2016
1
CHAPTER 1
1.2.2 GT-POST
Once the simulation is run, results may be sent to GT-POST, a graphical interface that enables viewing
and manipulating of the data collected from the simulation. GT-POST contains the following functions
and more:
Graphically based data post-processing
Quick generation of 2-D and 3-D plots of any quantity
Ability to combine data from different cases, different simulations or from tests into one plot
Re-usable post-processing templates allow instant cross-plotting of results from different
simulations
Ability to read data directly from ASC/Excel files with database capabilities for fast and flexible
access to measurement data.
Import or Export data to/from ASCII or MS/EXCEL files
Various math operations that can be performed on data in all plots (sum, difference, integral,
derivative, FFT…)
Animation of transient processes
User-definable tabular output of computed results
1.2.3 DOE Design Tool, DOE-POST (Design of Experiments), and Distributed Computing
With the Design of Experiments feature, many test cases may be run simultaneously (via built-in
Distributed Computing feature) and the user can define certain factors they wish to optimize, based on a
set of independent parameters. Several different DOE methods are available, including full-factorial and
Latin-Hypercube.
When the simulations are complete, DOE-POST can be used to find input values (independent variables)
that target or optimize one or more result variables. Independent variables maybe be "case independent"
(i.e. cam timing for a variable-valve-timing (VVT) engine) or fixed-sweep for all operating conditions
(i.e. cam-timing for a fixed-timing engine). User-defined weighting functions are available both for the
balance of result variables (when optimizing two or more values simultaneously) as well as for weighting
different operating conditions (e.g. more weight maybe be given to rated load operating point than to say
idle).
GT-SUITE's COOL3D preprocessor 3D flow-space solver differs from traditional full 3D CFD
approaches in two significant ways:
GT-SUITE typically runs much faster than a full 3D CFD solution because it is using a 1D
solution on the boundaries between subvolumes, while utilizing quasi-3D components (3D
flowsplit volumes, same as used to predict acoustic results in mufflers). GT-SUITE is mostly
complimentary to full 3D CFD in that the detailed analysis is very useful for more detailed spatial
resolution of the flow and thermal solution, such as to analyze localized temperatures (say surface
temperature of battery). But on the other hand, full 3D CFD is too slow to model many operating
conditions or to run Design of Experiments, where as COOL3D is practical for these applications
while producing equal results as 3D CFD in overall system performance such as mass flow rate
through a radiator or intercooler and sensitivity to spacing of thermal components.
GT-SUITE uses a discretized solution of the internal and external side of the heat exchangers to
capture interactions between the external underhood flow and the coolant (internal) flow side (for
radiators) or air (for intercoolers), typically being integrated with the cooling system.
GT-SUITE's COOL3D preprocessor and flow solver also differ significantly from other 0D/1D analysis
tools in that it is much more predictive. Other tools often require user input to impose (i.e. non-
predictive) flow distribution in regions where the flow is expected to re-direct around structural elements
or when flow redistributes due to pressure gradients, say due to variations in fan speeds between the
center and the blade tips, where as GT-SUITE predicts these differences.
Huge mode need not be used unless the user encounters a memory problem in a graphical application.
The application will notify the user with a warning message when this occurs. Some examples of error
messages are "Java Heap Space Error" or "Out of Memory".
To use the Huge Mode, there are two options. The first option is to temporarily run in Huge Mode. This
is the recommended option, because when the application is closed, the allocated memory will reset to the
default settings. However, if very large models or results files will be handled frequently, then the second
option is to permanently run in Huge Mode. The second option is not recommended unless the
workstation has at least 2 GB or more of RAM.
The first option is to start the application from a command line and append " -huge" after the executable
filename. To do this in Windows, create a command line by going to Start -> Run -> type "cmd" and
press the enter key, and then type "gtise -huge" and press the enter key. Please note that there is a single
space between "gtise" and "-huge". Likewise for GT-POST and COOL3D the commands are "gtpost -
huge" and "cool3d -huge", respectively. Depending on how your software is installed, you may need to
change the directory to %GTIHOME%\vX.X.X\GTsuite\bin\win32 for the executable filename to be
recognized (%GTIHOME% is usually C:\GTI).
The second option is to change the shortcuts to the GUI applications so they will always start in Huge
Mode. The shortcut can be either a shortcut icon on the desktop or the shortcut link in the programs list.
Please note that shortcut links to GTise Huge and GTpost Huge will be automatically added to the
programs list during installation. To change the shortcut link in Windows go to Start -> Programs -> GTI
Applications Group -> move the mouse pointer over the application in the list to highlight -> right-click
with the mouse -> select Properties -> in the Target box append " -huge" after the *.exe application
executable filename. Then press the OK button and start the program. Again please note that there is a
single space between *.exe and -huge.
If the Huge Mode setting is not large enough, the value for the maximum memory allocation can be
increased in the file called product.cfg. Note that depending on the system configuration, administrator
rights may be needed to modify the product.cfg file. Also note that the product.cfg file is overwritten
when a Build update is installed. To change the maximum memory allocation value:
Browse to %GTIHOME%\vX.X.X\GTsuite\bin
Make a backup copy of the file called product.cfg and save with a different name
Open product.cfg with a text editor
Search for "[GUI-all huge]". On the following line is a variable *VM_OPTIONS for controlling
the huge mode memory allocation. Change the number following Xmx in the VM_OPTIONS
string in order to change the maximum memory allocation.
The default is 1024 MB. A general recommendation is to increase in increments of 32 MB until the
application is able to function correctly. Larger increments, for example 128 MB, may be used if the
workstation has at least 2 GB or more of RAM. Be cautious when increasing the value. If it is set near or
above the total RAM of the computer, you may not be able to run other programs and/or the OS may
freeze/crash when trying to run the GUI application.
Many of the tasks listed here are not sequential; the user can go back and forth and change objects, parts,
run settings, and much more in between each run to vary the results as they please.
1.4.3 Tutorials
Perhaps the fastest way to become proficient with the software is to do the Tutorials. The tutorials
include both an instruction set in PDF file format and model files to accompany the exercises These
tutorials show the step-by-step process behind model building and focus more on answering the "How"
behind building and running simulations. The tutorial files can be accessed on the File -> Tutorials menu
or can be found in the installation directory at \GTI\v#.#.#\Tutorials
There are tutorials on almost every operation within GT-SUITE, and multiple exercises within each
tutorial. These tutorials are separated into three main groups: "Graphical Applications", "Modeling
Applications", and "Co-Simulation and User Routines". Tutorials within "Graphical Applications"
include GT-ISE, GEM3D, and GT-POST. The "Modeling Applications" tutorials cover basic to
advanced model building and manipulation of specific technologies from "Engine Performance" to
"Cooling and Thermal Management" to "Vehicle Driveline and Hybrid Electric Vehicles". "Co-
Simulation and User Routines" covers operations such as coupling GT-SUITE to CFD software or
Simulink.
1.6 Examples
There are many example models that are available to view, run, or even use as a starting point for specific
simulations. These examples can be found at %GTIHOME%\GTI\v#.#.#\examples (where
%GTIHOME% is the directory into which the software was installed). They can also be accessed easily
through GT-ISE, under "File" -> Examples. To view specific information on these examples, see the
Application-Specific User's Manuals. Each manual describes their corresponding examples in detail.
number, such as v7.1.0 to v7.2.0, etc. When a new version is released, it is sent to active lessees of GT-
SUITE, and so no request needs to be submitted to receive new versions.
New versions contain improvements to the code, which may cause a modest change in the results, but
always with the expectation that the change will make the results better. All changes which may change
results are carefully tested and evaluated before being introduced. Improvements which are expected to
significantly change results are typically implemented in a manner which allows the user to choose
whether or not to use the new method. The run time messaging is also modified such that the user is
made aware of the improved solution method. Because there are many factors and interdependencies that
affect results, there is no guarantee that the new version will necessarily correlate better to your
measurements in every model. The expectation is that if the correlation changes for a model, it is because
the new improvements exposed other issues that affect correlation.
GT-SUITE "Builds" are analogous to "Service Packs" released by other software companies. A build
might contain a new feature which is too urgent to wait for a new version or a bug fix. Builds are made
available approximately every three months, and are available for download (see www.gtisoft.com) to
active lessees of the software.
We strive to not introduce results changes between builds, even if the change would have corrected a bug.
It would be undesirable to allow results to change between builds as the user would not know if the
change was due to a change in the model (e.g. a dimension was changed) or due to software change.
Therefore, any change that might influence results will be added in a manner which allows the user to
choose to use the improvement. Furthermore the user is able to specify which build of the solver to use
for a simulation. This guarantees complete reproducibility of results from a build and should allow users
to update to the latest build without hesitation. See "Run with a specific build solver" below.
GT-SUITE includes several important features to help users to easily migrate to the newest version/build
to take advantage of newer features while simultaneously keeping continuity in results.
1) Multiple versions of GT-SUITE – The installation directory and licensing have been purposely
designed to allow more than one version of the software to be run at once. Simply install the new
version into the same folder as the old version(s). There will be separate icons to launch each
version of the software independently. Also, only one user-interface license will be utilized if a
user simultaneously runs a GT graphical application from multiple versions. GT recommends
that projects in progress should continue within the same version to maintain continuity of results.
New developments and those projects with an opportunity to transition should use the newest
version.
2) Reproduce results from the previous version – GT-SUITE is able to reproduce the results of
the previous version. This option is useful when a project needs to utilize a feature in the new
version, but continuity with previous results is also still necessary. This option is typically quite
reliable for simulations that are dominated by the flow solution (i.e. engine performance, cooling
systems, fuel injection systems, etc.), but is not supported by the mechanical applications. This
option is presented when evolving a model and can be changed in the folder under Setup
Advanced Setup VersionChanges. When choosing to run as an older version it is
recommended to compare results from the old to the new version before continuing the project in
the new version.
3) Specify the build of the solver – This was done to allow a user to update to the latest build with
the guarantee that results from a previous build can be exactly reproduced. Before using this
feature, the archived solvers must be installed on the machine(s) where the simulation will be run
(local/remote/distributed). This can be done by invoking the option when updating GT-SUITE
via gt-update. The archived solvers are not available when installing from the CD due to size
limitations. Then the solver build number is specified in the "Run Simulation Wizard" or in the
pull down menus File Options Run "Solver Build Number".
Also, future development of GT-SUITE is driven by daily support interactions with customers, from
whom valuable feedback and requests for both cutting-edge features come and simple usability
enhancements are received. New features are often developed in cooperation with customers, who
participate in their design, provide data for validation, and get involved in beta testing.
Contact Gamma Technologies Support by email at support@gtisoft.com (with your model attached, if
possible), or call during regular business hours at +1(630)325-5848, from 8 a.m. to 5 p.m. Central
Standard Time (GMT-6.00).
Plot Setup is useful when wishing to view all of the currently enabled plots and their
respective options. Within Plot Setup, one may choose to modify various plot properties,
including the location within a pipe where the plotted value will be measured, the plot range, x-
axis and sampling interval.
Gamma Technologies © 2016
11
CHAPTER 2
Alternatively, Case Setup also has provision to point directly to an ASC or EXCEL files. To define the
values for a parameter via this method, right click on the parameter in Case Setup, and then move it to an
Excel or ASC folder. (See Case Setup on-line help for more details.)
Before the solver is called, GT-ISE will perform some validation checks, and if errors are found, it will
issue messages in a warning box so that modifications can be made to the model. Double-clicking on any
of the blue warnings will open the appropriate dialog box to correct the mistake. GT-ISE will also
perform some checks on the estimated size of the output files, and if the estimated size is unusually large,
it will open a dialog box with options to reduce the output file size.
There are important Run settings available in GT-ISE, found in File Options Run that will determine
what occurs when the simulation is started (see figure below).
Depending on the "Show Simulation Wizard" checkbox setting at the top, either a "wizard" with
simulation options will pop up or else the simulation will launch directly (without prompt) always using
the settings given in the "Default Simulation Location" portion of the dialog box. Simulation options are
shown below. More detail about each of the options can be found in the help, accessed by clicking on the
icon in the top left corner of the simulation wizard.
If the Run Simulation button is disabled for the current *.gtm file, then either (1) the model is an external
subassembly and therefore cannot be run from this model file or (2) a solver window is already open for
the current model.
The new solver user interface divides numeric and non-numeric data into individual "scoreboards." Each
of the panes acts as dockable panels that can be moved around and resized within the solver UI. Each
scoreboard also has additional controls in the upper right hand corner to maximize, close, and reorganize
the data contained in the scoreboard, if applicable.
The scoreboards are defined in the "ScoreboardRLTs" and "UserScoreboardRLTs" folders found in
Output Setup. The "ScoreboardRLTs" folder contains all the "default" GT scoreboards. These
scoreboards will be automatically populated with data from the appropriate templates at runtime without
additional input needed from the user. However, the user has the ability to redefine which RLTs will be
reported for each template. Please see the template reference for 'RLTScoreboardListDef' for more
information on default scoreboards. The "UserScoreboardRLTs" folder is reserved for user-created
scoreboards. These are collections of RLTs from a given project map which are to be displayed at
runtime.
The "Messages" scoreboard in the solver user interface displays the same messages as the standard
console simulation window (with no graphical user interface) would. The double arrow in the bottom-
right corner of the message window toggles automatic scrolling of simulations messages on and off.
Manually moving the vertical scroll bar will also freeze scrolling. Bumping the scroll bar against the
bottom of its travel will also resume message scrolling. This is useful for reviewing the simulation status
or pertinent results as it is running.
The toolbar on the top allows the user to interact with the running simulation. Several buttons in the
toolbar have multiple options and these preferences are saved between simulations. For example, you can
immediately launch GT-POST to post-process your results from the simulation but the option to leave the
solver UI open or closed upon launching GT-POST can be selected via the drop-down arrow under the
button. This selection will be remembered for future runs.
Note to Linux users: To run a simulation without the solver UI enabled, you must have xterm installed.
If you do not have xterm installed, you must enable the solver UI. The solver UI may be enabled from
either GT-ISE or by using the "-g:on" option in command line (see the section "Running a Simulation
from a Command Prompt" for command line instructions).
The "Jump to Messaging for Case" button allows the user to find the beginning of each completed case in
the messaging output. Pressing the button opens a dialog informing the user which of the cases have been
completed and which one the user is seeking. The "Toggle Window Visibility" button opens a drop-down
list of all the scoreboards for the current simulation where they can be dismissed and recalled.
Alternatively, a scoreboard can be dismissed by pressing the red "X" in the upper-right-hand corner of
each scoreboard.
The layout, size, and organization of the solver UI can be retained for use in the future. The user can save,
load, and modify a perspective of their liking via the "Manage Perspectives" button. The named
perspective can then be recalled at a future date and the initialization of the UI can be changed from the
default GT method to the user's previous layout. Scoreboards will be recalled into a saved perspective
based on their names.
Note about Licensing: Running a simulation with the graphical user interface uses a GUI (Graphical
User Interface) license, in addition to the solver license. However, this should not affect the license usage
for most users because all graphical applications in GT-SUITE, for a given user on a given workstation,
share a single GUI license. For example, a user can simultaneously be running GT-ISE, GT-POST,
GEM3D, and two simulations with user interface, and still only use (i.e. share) one GUI license.
Considering that most all simulations are started from the GT-ISE user interface, the user will already
have a GUI license checked out, and thus the Solver UI will share that that license rather than checking
out an or additional license. In the rare instance when a GUI license needs to be freed while a simulation
continues, the display mode may optionally be switched from the user interface to a traditional messaging
window using the button in the Solver UI toolbar. (Please note that this will only be effective if all other
graphical applications in GT-SUITE are also closed.)
The central Distribution Server has another important benefit of providing a convenient way for a pool of
several users to share licenses, thus enabling the most efficient licensing usage. This server is aware of all
available processors and licenses, and can efficiently distribute jobs among the different workstations, as
well as provide a common queue for all users when not enough licenses are immediately available.
All of the files necessary for distributed computing are installed when GT-SUITE is installed, however
there are some steps necessary to configure distributed computing to run. For basic instructions on
configuring distributed computing, please see:
\GTI\v#.#.#\documents\Distributed_Computing\DistributedComputing.pdf.
This option has become less popular because the Distributed Computing feature described in the previous
section satisfies this same purpose, but with also much more capability. But one advantage of the Run
Remote option is that it is very simple, functioning very similarly to the Run Local option described
above. For example, when the simulation is launched via the Run Remote option, a command line
window will pop up, looking almost identical to the command line window for the Run Local option.
Note that the Graphical User Interface cannot be activated for a remote simulation. When the simulation
is complete (either because it finished successfully or because an error message has been issued), the files
are automatically copied back to the local workstation. If the window is closed before the simulation has
completed, the run will be terminated, just like when run locally. If GT-ISE is closed while a simulation
is running remotely, then the simulation will continue to run on the remote workstation, however, the
screen output will close and the result files will have to be manually copied back to the workstation.
Therefore, GT-ISE should typically be left open when using this option.
Remote runs can occur between any two networked workstations regardless of platform and operating
system. The "remote workstation", also known as a "solver node", must first be configured to run as a
Solver Node (i.e. to accept simulations from other workstations). Once a workstation has been configured
as a remote solver node, it can receive simulations from both Run Remote option and from Distribution
Server (see Distributed Computing below). Please see
\GTI\v#.#.#\documents\Distributed_Computing\DistributedComputing.pdf for details about setting up a
solver node
If a Remote Solver Node is available, GT-ISE can be configured to find the solver node in File > Options
> Run Remote. This includes the remote hostname, username, and port number (usually 3490).
Note that it is also possible to run remotely from command line. This is discussed later in this chapter in
the section about running from command line under the "-remote" option.
It is recommended to pre-process files that are to be run in batch mode so that any errors are discovered at
the time the simulations are submitted (instead of several hours later when the simulation starts to run,
when you may be away from your desk).
When Run Batch is started, all files that are open in GT-ISE that are not already queued will appear in the
left window and all files in the queue appear in the right window. The batch queue lists the order in
which the simulations will be run. The "Processed" tab in the right window will show files that are
finished along with their status (i.e. successful or error).
To move a file into the batch queue, first select the file by highlighting it in the left window and then use
the >> button. The first file transferred to the queue appears on top and any additional files appear below
this file. The files are run in the order shown on queue listing, from top to bottom. To change the order
of the files, use the Move Up or Move Down. Files may be removed from the queue by using the
Remove button.
The PAUSE option allows a break to be place in the batch file. Pause does not interrupt the simulation of
the files above it in the queue, but rather stops that batch from continuing on to the next file after the
pause. This option can be useful when a user wants to run some subset of the total queued files. The
PAUSE option may be moved into the batch queue in the same manner as files queued.
If it is desired to create a queue of simulations but run them later, choose and order the files as desired,
and then hit the Save button at the bottom. Simply reopen this dialog as necessary to add more files or
and modify the list as desired before starting the batch run. This option can be a convenient way to
prepare simulations to run overnight as work is being done throughout the day.
When the desired files are in the batch queue and in the correct order, select the Start Batch button. Once
a model is running, its Status will change from "Queued" to "Running". Once a simulation is complete,
the Status will change to "Completed". It is possible to add files to, change the order of or delete files
from the queue while other simulations are running, with the exception of the file(s) running at that
moment.
"Stop Batch" may be selected while running in batch mode to "Halt" the batch run. This will cause the
file(s) that are currently running to finish the current cycle, run one more cycle then stop and save the
results. The remaining cases will not be run. However, the queue will not be cleared. The remaining
files in the batch queue may be restarted by returning to the Simulation Batch Setup, removing the
PAUSE in the queue list and selecting Start Batch.
Please note that the regular "Halt/Suspend Simulation" on the Run menu and toolbar may also be used to
halt the current run as desired as well, but unlike the "Stop Batch" button, this will not insert a PAUSE
into the batch queue. Halting the current case and suspending the current run are not available for batch
runs. Also, please note that the new graphical user interface is not available during batch simulations.
Execution of a batch simulation creates a file called "gtbatch.log" in a temporary directory. This file can
be viewed by clicking on the button "Log file…" the Simulation Batch Setup window. It will be shown in
a window and the location of the file will be displayed on the title bar of the window. In addition, the file
can with any text editor (e.g. Notepad, Jot, vi, etc.). The log file provides a summary of the commands
that were executed and reports any errors or problems that may have occurred during the batch
simulation.
Halting a simulation can be done from the GT-ISE toolbar or from the new graphical interface. Also, it
can be done without any interface. This is accomplished by creating a modelname.hlt text file where
modelname is the name of the model that is running that needs to be halted. The text inside the hlt file will
determine which halt option is invoked. Please see the table below for more details.
The dialog above can also be used to pause (or suspend) a run. To resume the suspended run, click on the
simulation output window and press any key. To suspend the simulation without using the graphical
interface, a modelname.hlt file must be created with the text "suspend" in the file.
An alternative option exists to resume the run based on the existence of a modelname.rst file (with no text
in it). To use this resume method, the environment variable gtsuspend must be defined at the time the
simulation was started.
In some situations, it may be desirable to run a simulation from a command line. It is not recommended
to call the solver executable directly. Rather, command line runs should use the script (batch) files
provided in the $GTIHOME\bin directory (on Windows this is typically C:\GTI\bin). This directory is
added to the path during installation, and so the commands should be readily available without providing
a path to the script. To start the simulation from the prompt, type:
where filename is the name of the file you wish to simulate and [flags] includes any desired optional
flags. The [optional flags] argument controls many things from minor details like displaying
monitors, to running the file remotely on the distributed queuing system. To see the complete list of
options available type "gtsuite -h" at the command line.
The "gtsuite" command should always be used to invoke the solver, regardless of the file's project/license
type (i.e. GT-SUITE, GT-POWER, GT-SUITEmp, GT-POWERLab, etc.), as the license type will
automatically be determined from the information in the .gtm file.
As part of the process, a *.dat file, which is the file that the solver literally reads, will automatically be
created. If a .dat file of the same name already exists, it will be overwritten. Though it is possible to
directly specify to a .dat file directly from the gtsuite command as the filename, this is not recommended
as some features are not fully supported in this configuration, and thus unexpected (and undesirable)
results may occur.
These are three ways used to select the software version that will be used, in order of precedence:
2. If a *.gtm file is run, the version is automatically determined from information within the file.
3. If the file is not .gtm, and the "–V version " command line argument is not given, a small text file
named VERSIONS.TXT is used. VERSIONS.TXT resides in the $GTIHOME directory (i.e.
c:\GTI), and lists a version number for each application. This file is overwritten each time a new
software version is installed, so it should list the most recently installed version numbers, unless
otherwise modified by the user. The text file can be edited to change the version number if
desired.
Sometimes several different users will share a common installation directory from a network server. In
this situation, it is desirable to allow each user to have their own VERSIONS.TXT file outside of the
common installation directory. The user may override the settings in the installation directory
(/GTI/VERSIONS.TXT) by copying the VERSIONS.TXT file into their home directory. The home
directory is defined as follows:
For Windows, the home directory is defined by environment variable USERPROFILE. By
default, USERPROFILE is set to c:\Documents and Settings\username, where username is the
login name of the user. To verify the setting of USERPROFILE, type "SET" from a Command
Prompt window.
For Linux, the home directory is the simply the standard defined home directory (~).
This method is ideally suited for a situation where the optimization task is well defined and involves only
a single result variable RLT that can be optimized independently for each simulation case. For example,
this method could be used to determine the exact diameter of an orifice that creates a specific upstream
pressure. The reason that this method is preferred for this type of situation is that it will generally result
in the minimum number of solver iterations to produce the desired response. This is because the solver
uses an algorithm to logically determine the input values for the next iteration based on feedback from
results of prior iterations. However, this algorithm does not make the direct optimizer a good selection
when the overall shape of the RLT response surface is of interest. The algorithm is concerned only with
determining the optimal point, and thus there may be very little data in the areas of the input design space
that are not near the optimal point. This may make it impossible to draw conclusions about the
relationship between the input parameters and the result variable in these other areas.
The feedback algorithm that is used for the direct optimizer also has other implications to consider. If the
setup criteria for the optimization must be changed, or if it is necessary to optimize a different RLT
variable, it is necessary with the direct optimization method to set up and run a new optimization (this is
not necessarily the case with the design of experiments method). In addition, because the solver iterations
are performed within each simulation case, the iterations cannot be distributed among different
processors.
The setup for the direct optimizer is found in GT-ISE under the Optimization -> Direct Optimizer. For
more details on the specific attributes within the optimizer, please click on the on-line help button within
the optimizer dialog.
processing environment to fit a mathematical surface through the data for one or more RLT variables.
Once this response surface has been established, it is possible to quickly and easily perform optimizations
on this surface. In this method, the criteria for the optimization are entered into the post-processing
environment where the optimization will take place. The GT-SUITE solver in this case has no knowledge
of the eventual optimization task, and simply runs a predefined set of experiments. The purpose of the
solver is simply to generate the data over the complete range of inputs that will be used to define the
response surface in the post processor.
This method will typically require more points to be run than the direct optimization, and requires post
processing of results to perform the actual optimization. However, this method has several benefits that
are gained because there is no feedback mechanism within the runs as in the direct optimizer. First,
because each experiment is a separate independent case, it is possible to use the distributed computing
feature to divide up the runs among different processors. This can actually make the total runtime for this
method less than the total runtime for the direct method, even though more points may need to be run.
Second, multiple different optimizations may be run in the post-processing environment without re-
running any simulations. This is because the original set of experiments that were run covered the whole
input design space. Once the resulting RLT data is loaded into the post processor and the response
surface created, it is possible to run as many different optimizations on the surface as desired, including
different result RLT variables. It is also possible to load in multiple RLT variables to the post-processor
at the same time if the optimization task requires the optimization of an objective function that is a
combination of two or more RLT variables.
The DOE Post-processor also makes it possible to run a "sweep" optimization. This is an optimization
where the values of the input parameters must be identical for all simulation cases considered, and the
optimal value of the result RLT is a weighted average across all of the cases. For example, in an engine
with a fixed geometry camshaft, the intake valve timing will be the same for all engine speeds. In this
case, it may be desirable to optimize the timing value by considering the volumetric efficiency of the
engine across all engine speeds. The timing value determined by the post-processor in a sweep
optimization will not necessarily be the optimal value at each speed when considered independently, but
will provide the best overall efficiency across all engine speeds according to the weighted average. The
user may weight cases differently to make some more important than others.
To setup a DOE, go to DOE Setup... under the Optimization menu in GT-ISE. Additional details on
DOE Setup can be found in the on-line help within that dialog. Once the DOE simulation has been run,
the DOE Post Processor can be launched by selecting Launch DOE Post... from the Optimization menu.
More details in the DOE Post-Processor can be found in an Appendix of this manual.
1) Configure the model .gtm file in GT-ISE to have one case in Case Setup.
2) Make sure that each attribute that is to be optimized is defined as a parameter.
3) Make the particular parameters of interest to the optimizer read from an external file, such as (a)
External parameter file (see next section) (b) ASC or EXCEL file (see Folders options in Case
Setup--refer to Case Setup on-line help for more detail).
4) Configure the external optimizer to write out desired input variable(s) for each case to the
external file.
5) Run simulation from the command line (see section above)
6) Extract result(s) of interest from the result files using the "gtexport" command line utility. For
more details on this utility, please see the GT-POST manual, which is available from the Help
menu in GT-ISE or GT-POST (File > Manuals > Graphical_Applications > GT-POST.pdf)
7) If more iterations are required, create a new external parameter file with new input values. Go
back to step 3 and repeat as necessary until the study is finished.
When referring to an external file from GT-ISE, the easiest method is to use the Import Wizard to select
the file and data range. The Import Wizard can be started by clicking on the gray box with three dots next
to an array attribute and selecting the button named '<Filename>'. After the wizard is complete, the
filename and data range will appear in <brackets>. The path of the file may optionally be included with
the filename. However, the path is not necessary if the file is located in an alternate directory. GT-
SUITE will look in other directories for external file references in the order shown below. If the file is
not located in any of these directories then it will not be found.
1. Optional path specified in <brackets>.
2. Local directory where model is located.
3. Directories specified in the Paths folder of Advanced Setup. These will be looked at in the order
they are input.
4. Directories specified with the Path environment variable.
If the file is a text file, it can be created by any text editor or program that can save an ASCII text file,
such as Notepad in Windows or VI in UNIX. In some editors, and in almost all word processors, the
"File Type" option in the "Save As" dialog box will need to be used to prevent additional formatting from
being stored. A number of delimiters are supported using the Excel/ASCII Import Wizard. The default is
tab or space delimited, but the wizard supports a wide range of delimiters using the check boxes for
semicolon, comma, and "Other". The "Other" field will accept any single character as a delimiter. Multi-
character delimiters are not supported.
GT-ISE. These attributes can point at such data if the data is prepared in either of the following two
formats:
The data may be arranged in columns. This is often convenient for data that already exists in a
spreadsheet. To point to an external text file, click on the Value Selector for the first row in the
column. Click <File Name> in the upper right-hand corner to open the Import Wizard. This wizard
will guide you through selecting your file, delimiters, and data ranges much like the Microsoft Excel
text file importer. Once the wizard is completed, the following syntax will appear in the cell:
<filename#StartColumn#EndColumn#StartRow#EndRow#delimiters>
Example: <data.txt#0#0#0#-100>
where the filename is the name of the file including extension and the path (optional), StartColumn is
the first column of data to use in the text file and EndColumn is the last column of data of interest.
StartRow is the first row of data to use in the text file and EndRow is the last row of data of interest.
Note that both row and column numbers start with 0 in this format. The number "-100" indicates that
the file should be read until the end of the data is reached. The delimiters flag specifies any optional,
non-default delimiters selected when using the Import Wizard. With this format it is not necessary to
modify the text file in any way such as rearranging labels as the references may be used to extract
exactly the data required.
The data may be placed in blocks within the file. Each block should start with the short variable
name of the attribute followed by the ">" symbol. The data should then follow, separated by spaces
(never use tabs); the data may be in either integer or real-number format. The data may be put on
several lines if the user desires. At the conclusion of each data set, a "<" symbol must be used to
indicate the last data in that set has been read. Any number of data blocks may be listed in a single
text file as long as each data block has a unique short variable name. Below is a short example of two
arrays from the 'ValveCam' reference object that could be used to define the valve lift. If the user
would like to input the "THET" data into the column with a short variable name of "THET", then
THET> would be input into the attribute cell.
To point to an Excel file from an array attribute, click on the Value Selector for the first row in the
column. Click <File Name> in the upper right-hand corner to open the Import Wizard. This wizard
will guide you through selecting your file and data ranges. Once the wizard is completed, the
following syntax will appear in the cell:
<filename#SheetName#StartColumn#EndColumn#StartRow#EndRow >
Example: <data.xlsx#Sheet1#1#1#0#-100>
where the filename is the name of the file including extension and the path (optional), SheetName is
the name of the sheet in the Excel file to read from (this name is case sensitive), StartColumn is the
first column of data to use in the text file, EndColumn is the last column of data of interest, StartRow
is the first row of data to use in the text file, and EndRow is the last row of data of interest. Note that
both row and column numbers start with 0 in this format. The number "-100" indicates that the file
should be read until the end of the data is reached. If the StartColumn and EndColumn are different,
then the data is read across rows first, then columns (like block or variable format text files above).
For example, if StartColumn and StartRow were A1 and EndColumn and EndRow were C3, then the
data would be read in A1, A2, A3, B1, B2, B3, C1, C2, C3.
When referencing data from a .gdx/gdt file, the Import Wizard lets the user pick from three different data
type: Instantaneous Plots, Case RLT and Time RLT. Depending on the data type, the Import Wizard
helps the user select the data and it will construct the syntax that will appear in the cell. For instantaneous
plots, the data index name has to be obtained from GT-POST but for Case RLT and Time RLT, the RLT
Selector is available to help the user select the RLT of interest.
To point to a .gdx/.gdt file from an array attribute, click on the Value Selector for the first row in the
column. Click <File Name> in the upper right-hand corner to open the Import Wizard. This wizard will
guide you through selecting your file and data. The screenshot of the Import Wizard used for .gdx/.gdt
data import is shown below.
Instantaneous Plots: The data set must be of type XY. For instantaneous plots, 'Data Index' name has
to be entered in the 'Data Index Name' field of the Import Wizard. 'Data Index' can be found by
opening the *.gdx/*.gx file associated with the data in GT-POST, editing the "Properties" of the data
set that is of interest, and the value will be shown under the Source tab "Index" field. The user needs
to copy the "Data Index" from GT-POST and paste it into the 'Data Index Name' field in GT-
ISE\Import Wizard. Once the wizard is completed, the following syntax will appear in the cell:
<filename#PlotArray#DataIndex#CaseNumber>
where the filename is the name of the file including extension and the path (optional), PlotArray
indicates whether it is the X-data or Y-data, and CaseNumer is the case number
Case RLT: The "Select RLT" button in the Import Wizard can be used to select the RLT of interest.
By clicking on this "Select RLT" button, the RLT Selector will come up so that the user can select the
RLT. Once the wizard is completed, the following syntax will appear in the cell:
Gamma Technologies © 2016
28
CHAPTER 4
<filename#CRLT#RLTIndex>
where the filename is the name of the file including extension and the path (optional), CRLT indicates
that this is a Case RLT data set.
Time RLT: The "Select RLT" button in the Import Wizard can be used to select the RLT of interest.
By clicking on this "Select RLT" button, the RLT Selector will come up so that the user can select the
RLT. Once the wizard is completed, the following syntax will appear in the cell:
<filename#TRLT#DataIndex#CaseNumber>
where the filename is the name of the file including extension and the path (optional), TRLT indicates
that this is a Time RLT data set, and CaseNumer is the case number.
Please note that this feature only works with the 32-bit version of GT-SUITE applications in v2016
Build1 but it will be functional with 64-bit of GT-SUITE starting Build2 of v2016. If a user tries to use
this feature with the 64-bit version in v2016 Build1, an option to open the 32-bit version of GT-ISE will
be provided as shown below.
After selecting a MDF file, all the Channel Groups stored in the MDF file will be displayed as shown
below. By expanding the Channel Group tree, data sets can be selected and the preview of the data is also
available to the user to check the values before importing them.
This functionality is also available in GT-POST. Using a Report (.gu) file, a user can add any external
data by right-clicking on a plot and selecting "Add Data…" as shown below.
The Add Data Wizard in GT-POST will show an option to use MDF file as the data source as shown
below.
After selecting a MDF file, a user can see all the channels on the left side of the wizard. Drag and drop
the channels from left pane to the right pane as shown below.
On the next page of the wizard, X and Y should be assigned to the selected channels. After hitting
"Finish", the data from the MDF file will be added to the Report (.gu) file.
A summary of each example has been included in each model to describe the purpose of the model, how
features and templates are being used, and what modeling or design function is being achieved.
To see details of what each command can do, please review the help text by typing "-h" after the
command shown below, unless otherwise noted.
Commands:
converge_post - Converts CONVERGE-Lite output files to formats compatible with 3rd
party software (additional command line help is not available)
cool3d - Launches COOL3D
cool3dc - COOL3D console - Used for automating tasks
gem - Launches GEM
gemc - GEM console - Used for automating tasks
getid - Displays name of hardlock/dongle connected to computer.
gtadmin - Opens administrator settings for GT-ISE users (additional command line help is not
available, but help is available within application)
gtexport - Exports GT-POST data to a text file
Please see the section Automatically Export Data to Text File
DistributedComputing and the section Automatically Export Data
-POST for more information on data export.
gtise - Launches GT-ISE
gtisec - GT-ISE console - Used for automating tasks
gtlink - Launches solver in server mode for CAE coupling simulations
gtpost - Launches GT-POST
gtsuite - Launches the solver.
o gtpower, gtpowerx, gtsuite, gtsuitemp are aliases for to gtsuite to start the
solver. The gtsuite command can be used to start the solver regardless of license
used.
o For detailed information on running the solver from command line, please refer to
Section 2.4 of this document, Running a Simulation from a Command Prompt
(Windows or Linux).
gtsuitert - Launches real-time solver
vtdesign - Launches VT DESIGN
INDEX
3 F
3D CAD files (GEM3D), 3 file references
external file formats, 25
A File Types, 7
arrays
G
.gdx files, 27, 29
Excel files, 26 gem, 34
external files, 25 GEM 3D, 3
ASCII files, 25 gemc, 34
getid, 34
B gtadmin, 34
gtexport, 34
batch files gtise, 34
command line, 20 GT-ISE, 2
batch simulations, 16, 17 gtisec, 34
builds, 8 gtlink, 34
gtpost, 34
C GT-POST, 2
CAD files (GEM3D), 3 GT-POWER, 1
cam design tool, 4 GT-POWERlab, 1
Case Setup, 11 gtsuite, 34
command line, 34 GT-SUITE
command line simulation, 20 examples, 7
converge_post, 34 training, 7
cool3d, 34 versions and builds, 8
COOL3D, 3 GT-SUITEmp, 1
cool3dc, 34 gtsuitert, 34
customer support, 10 GT-SUITE-RT, 2
D H
M remote simulations, 16
Run Remote, 16
manuals
Run Setup, 11
on-line help, 6
reference manual, 6
tutorials, 6 S
users manuals, 6 script files
memory allocation (Huge Mode), 4 command line, 20, 34
modeFrontier (optimization), 23 support, 10
Suspend simulation, 19
O
on-line help, 6 T
optimization, 22 text files, 25
direct, 22 training, 7
DOE, 3 tutorials, 6
DOE, 22
optimization, external software, 23 U
Output Setup, 11
underhood modeling (COOL3D), 3
P
V
parameters
changing via External File, 12 valvetrain design tool, 4
pause simulation, 19 versions, 8
Preprocess Simulations, 12 vtdesign, 34
VTDesign, 4
R
W
RAM (Huge Mode), 4
Red Cedar (optimization), 23 web update, 8