ANSYS Forte Best Practices

Forte Best Practices
ANSYS, Inc. Release 2019 R1

Southpointe January 2019
2600 ANSYS Drive
Canonsburg, PA 15317 ANSYS, Inc. and
ansysinfo@ansys.com ANSYS Europe,
Ltd. are UL
http://www.ansys.com registered ISO
(T) 724-746-3304 9001: 2015
(F) 724-514-9494 companies.
Copyright and Trademark Information
© 2018 ANSYS, Inc. Unauthorized use, distribution or duplication is prohibited.
ANSYS, ANSYS Workbench, AUTODYN, CFX, FLUENT and any and all ANSYS, Inc. brand, product, service and feature
names, logos and slogans are registered trademarks or trademarks of ANSYS, Inc. or its subsidiaries located in the
United States or other countries. ICEM CFD is a trademark used by ANSYS, Inc. under license. CFX is a trademark
of Sony Corporation in Japan. All other brand, product, service and feature names or trademarks are the property
of their respective owners. FLEXlm and FLEXnet are trademarks of Flexera Software LLC.
Disclaimer Notice
THIS ANSYS SOFTWARE PRODUCT AND PROGRAM DOCUMENTATION INCLUDE TRADE SECRETS AND ARE CONFID-
ENTIAL AND PROPRIETARY PRODUCTS OF ANSYS, INC., ITS SUBSIDIARIES, OR LICENSORS. The software products
and documentation are furnished by ANSYS, Inc., its subsidiaries, or affiliates under a software license agreement
that contains provisions concerning non-disclosure, copying, length and nature of use, compliance with exporting
laws, warranties, disclaimers, limitations of liability, and remedies, and other provisions. The software products
and documentation may be used, disclosed, transferred, or copied only in accordance with the terms and conditions
of that software license agreement.
ANSYS, Inc. and ANSYS Europe, Ltd. are UL registered ISO 9001: 2015 companies.
U.S. Government Rights
For U.S. Government users, except as specifically granted by the ANSYS, Inc. software license agreement, the use,
duplication, or disclosure by the United States Government is subject to restrictions stated in the ANSYS, Inc.
software license agreement and FAR 12.212 (for non-DOD licenses).
Third-Party Software
See the legal information in the product help files for the complete Legal Notice for ANSYS proprietary software
and third-party software. If you are unable to access the Legal Notice, contact ANSYS, Inc.
Published in the U.S.A.

Table of Contents
1. Forte Best Practices Introduction ........................................................................................................... 1
2. Best Practices .......................................................................................................................................... 3
2.1. Operating Environment .................................................................................................................... 3
2.2. DI Diesel Engine Case (Typically, Sector Mesh) ................................................................................... 3
2.2.1. Initial Set-up ............................................................................................................................ 3
2.2.1.1. Sector Mesh Generation .................................................................................................. 3
2.2.1.1.1. Bowl Profile ............................................................................................................ 3
2.2.1.1.2. Topology Selection, Mesh Control Points, and Size Parameters ................................. 4
2.2.1.2. Evaluating Initial (IVC) Gas Composition ........................................................................... 5
2.2.1.2.1. Initial (IVC) NOx for High EGR Cases ......................................................................... 5
2.2.1.3. Fuel Mass for Direct Injection Cases .................................................................................. 6
2.2.2. Basic Calibration Procedures ..................................................................................................... 6
2.2.2.1. IVC Pressure and Temperature ......................................................................................... 6
2.2.2.2. Injection Timing .............................................................................................................. 6
2.2.3. Advanced Calibration ............................................................................................................... 7
2.2.4. Comparing to Experimental Data .............................................................................................. 8
2.2.4.1. Calculating Apparent Heat Release Rate from Experimental Pressure ................................ 8
2.2.4.2. Importing Data Arrays into Forte Visualizer ....................................................................... 9
2.2.4.3. Review Spray Behavior ................................................................................................... 10
2.2.4.4. Emissions and Performance Data ................................................................................... 10
2.2.4.4.1. Conversion to Mass per Time Units ........................................................................ 11
2.3. Setting Up Automatic Mesh Generation Cases With Valve Motion ..................................................... 11
2.3.1. Read In and Process the Surface Mesh .................................................................................... 12
2.3.1.1. Selecting the Source of Surface Meshes ......................................................................... 12
2.3.1.2. Splitting the Geometry .................................................................................................. 12
2.3.1.3. Normals ........................................................................................................................ 14
2.3.2. Set Up Mesh Size and Mesh Refinement .................................................................................. 15
2.3.2.1. Material Point ................................................................................................................ 15
2.3.2.2. Global Mesh Size ........................................................................................................... 15
2.3.2.3. Mesh Refinement .......................................................................................................... 15
2.3.2.4. Modeling the Crevice ..................................................................................................... 20
2.3.2.5. Modeling Sector Geometry Using Automatic Mesh Generation ...................................... 20
2.3.2.6. Multi-Cylinder Simulations ............................................................................................. 21
2.3.3. Models ................................................................................................................................... 21
2.3.3.1. Turbulence Model .......................................................................................................... 21
2.3.3.2. Flame Speed Model for SI Engine Simulations ................................................................ 21
2.3.3.2.1. SAM Settings for Flame Propagation ...................................................................... 24
2.3.3.3. Spray Model .................................................................................................................. 24
2.3.3.3.1. SAM Settings for Spray Vaporization ...................................................................... 25
2.3.3.4. Spark Model .................................................................................................................. 26
2.3.3.5. Soot Model .................................................................................................................... 26
2.3.4. Boundary Conditions ............................................................................................................. 27
2.3.4.1. Boundary Profiles .......................................................................................................... 27
2.3.4.2. Valves ............................................................................................................................ 27
2.3.4.3. Valve Angles .................................................................................................................. 27
2.3.4.4. Valve Lift ....................................................................................................................... 28
2.3.4.5. Movement Type ............................................................................................................. 28
2.3.5. Initial Conditions .................................................................................................................... 29
2.3.5.1. Initialization Order ......................................................................................................... 30
2.3.5.2. Simulation Controls ....................................................................................................... 30
Release 2019 R1 - © ANSYS, Inc. All rights reserved. - Contains proprietary and confidential information
of ANSYS, Inc. and its subsidiaries and affiliates. iii
Best Practices
2.3.5.3. Output Controls ............................................................................................................ 32

2.3.5.4. Preview Simulation ........................................................................................................ 33
2.3.5.5. Forte Engine Simulation Checklist .................................................................................. 33
2.4. Setting Up Two-stroke Engine Cases ................................................................................................ 34
2.4.1. Piston/Cylinder Wall Gap ........................................................................................................ 34
2.4.2. Locating the Piston and Defining Motion ................................................................................ 37
2.4.3. Surface Mesh Topology and Other Considerations .................................................................. 39
2.4.4. Runtime Port Open/Close Considerations ............................................................................... 40
iv of ANSYS, Inc. and its subsidiaries and affiliates.
List of Figures
2.1. Sector mesh bowl profile and topology control points ............................................................................. 5
2.2. STL dialog when reading STL files into Forte .......................................................................................... 13
2.3. Green region is the valve, pink is the port. Note the donut surface around the intersection of the two ..... 14
2.4. Suggested refinements for typical engine cases ..................................................................................... 16
2.5. SAM control for resolving flow fields ...................................................................................................... 18
2.6. SAM control for resolving reactive regions ............................................................................................. 19
2.7. Turbulent flame-speed input parameters ............................................................................................... 22
2.8. Example of a Flame Library for a TRF gasoline ........................................................................................ 23
2.9. SAM settings for flame propagation ...................................................................................................... 24
2.10. Copy and Paste buttons located at the top of Editor panels .................................................................. 25
2.11. SAM settings for spray vaporization ..................................................................................................... 26
2.12. Valve stem split off from the valve face ................................................................................................ 27
2.13. Valve tilt angles ................................................................................................................................... 28
2.14. Indicating the marker point location for a region ................................................................................. 30
2.15. Turbulent Kinetics Interaction Model (TKI) panel .................................................................................. 31
2.16. Dynamic Adaptive Chemistry (DAC) panel ........................................................................................... 32
2.17. Sample geometry with gap demonstration .......................................................................................... 35
2.18. Gap too large ...................................................................................................................................... 35
2.19. Gap too small ..................................................................................................................................... 36
2.20. Acceptable gap ................................................................................................................................... 36
2.21. Mismatch in angular resolution causing self-intersections .................................................................... 37
2.22. BDC position translated to TDC ............................................................................................................ 38
2.23. Motion Type set as Sliding Interface ..................................................................................................... 39
2.24. Directions of normal vectors ............................................................................................................... 39
2.25. Geometry with enclosed piston ........................................................................................................... 40
2.26. Geometry with moveable piston ......................................................................................................... 40
of ANSYS, Inc. and its subsidiaries and affiliates. v
vi of ANSYS, Inc. and its subsidiaries and affiliates.
List of Tables
2.1. Parameters providing secondary impacts on engine behavior ................................................................. 7
of ANSYS, Inc. and its subsidiaries and affiliates. vii
viii of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 1: Forte Best Practices Introduction
ANSYS Forte best practices are presented in this book on these topics:
• Diesel engine simulation
• Automatic mesh generation (AMG)
• Two-stroke engine simulation
Further information about ANSYS Forte is available in the Forte Quick Start Guide, Forte User's Guide,
Forte Theory Manual, and Forte Tutorials. Documentation and tutorials are availalable from the ANSYS
Help site.
of ANSYS, Inc. and its subsidiaries and affiliates. 1
2 of ANSYS, Inc. and its subsidiaries and affiliates.
Chapter 2: Best Practices
This chapter presents the current best practice recommendations for ANSYS Forte, covering the following
topics:
2.1. Operating Environment
2.2. DI Diesel Engine Case (Typically, Sector Mesh)
2.3. Setting Up Automatic Mesh Generation Cases With Valve Motion
2.4. Setting Up Two-stroke Engine Cases
2.1. Operating Environment

For specific hardware and operating system specifications, see the Platform Support page at ansys.com.
Recommended configurations for ANSYS Forte installations:
• For chemistry-intensive simulations with a large mechanism (>300 species), we recommend 8–16 CPUs per
job.
• Processor: 64-bit with recommended memory. Intel is recommended over AMD. The faster the processor,
the better.
• Memory: 6 GB per core.
• Parallel operation: Intel MPI is required; MPICH2 may be compatible but is untested and is not officially
supported.
• A large hard disk is required for saving solution files.
2.2. DI Diesel Engine Case (Typically, Sector Mesh)

2.2.1. Initial Set-up
• Units:
Pay special attention to units. Be sure you enter units correctly as specified for the case, and do all
the conversions correctly or allow the Simulation Interface to convert values for you whenever possible.
2.2.1.1. Sector Mesh Generation
These comments apply to using the Sector Mesh Generator , which opens into its own window
from the Geometry Editor panel.
2.2.1.1.1. Bowl Profile

Obtain the bowl profile data. Sometimes these are given in the form of spline segments that have been
extracted from CAD drawings or another mesh. You must order these coordinate points and then convert
these to X, Z pairs. The X-Z coordinate frame must be defined as follows:
Best Practices
1. X is the radial direction from the center axis of the cylinder toward the cylinder wall (liner). A value of X=0
is at the cylinder center axis. At the cylinder wall, the value of X should be Bore/2.
2. Z is the vertical distance from the face of the piston lip in the direction of the cylinder head. The value of
Z=0 must be at the outer edge of the piston face (closest to the cylinder wall). Z coordinates into the piston
bowl will be negative.
3. The first point in the bowl profile should be the outer point on the piston face nearest to the cylinder wall
(Z=0, X= Bore/2 - crevice_width). If the X coordinate is smaller than Bore/2 - crevice_width, the mesh gener-
ator will automatically extend the profile to this point. The actual first point will be set to (Z=0, X= Bore/2
- crevice_width).
4. If you have no information about the crevice, start with the default crevice values (1 cm long, 0.1 cm wide).
Note that the “crevice width” cannot be set to 0, even if the Include crevice volume option is un-
checked. Similarly, the radial cell count corresponding to the crevice width cannot be 0. In other
words, even if you do not want to include the crevice block, the topological block above the crevice
must always be retained.
If you do not want to mesh the crevice volume, the Include crevice volume option should be un-
checked, and the Axial cell count corresponding to the crevice volume should be set to 0.
Only include points that describe the profile from the piston face outer edge to the tip of the bowl at
the center axis. Do not include points on the cylinder head.
The Sector Mesh Generator will assume a flat cylinder head. If this is not desired, you must generate
the mesh using ICEM CFD or K3PREP, or consider using a 360° mesh instead of a sector mesh. Forte
also is able to model sector engine geometry using Automatic Mesh Generation, which can also handle
sector geometry with a non-flat cylinder head.
2.2.1.1.2. Topology Selection, Mesh Control Points, and Size Parameters

Once you have defined and saved the profile, the next step is to determine which topology is most
appropriate for your profile. Look at the different topology options and select the one that looks the
most like yours.
Determining Mesh Control Points and Size Parameters is an iterative trial-and-error process. The Mesh
Control Point “fractions” are fractional distances along the entire length of the bowl profile. So if you
think about stretching out the profile into a straight line and scaling it to a distance of 1.0, the control-
point location parameters are the distance where each control point lies along this line, starting from
the outer edge (Z=0, X=Bore/2-crevice_width). Make your best guesses, since it is easy and quick to iterate
once you create an initial mesh. For example, for Topology #1, you might have 0.25,0.3,0.4, 0.7, for your
starting locations for Control Points 1,2,3,4, respectively.
Cell counts are estimated based on trying to make the side size of any cell somewhere between 1 and
2 mm, where ~1.3 would be better for production runs, but ~2 might be better for initial setup verific-
ation (that is, the coarser mesh will run faster).
DI Diesel Engine Case (Typically, Sector Mesh)
Figure 2.1: Sector mesh bowl profile and topology control points
2.2.1.2. Evaluating Initial (IVC) Gas Composition

Use the Composition Calculator utility to calculate the following compositions:
1. Intake port/inlet composition for engine cases with valves and ports.
2. Exhaust port initial composition for engines cases with valves and ports.
3. In-cylinder initial composition at IVC for engine cases that are in-cylinder only.
The Composition Calculator can be accessed from the Utility menu or the Composition Calculator
button on the toolbar. Please refer to the Forte User's Guide, for further details of how to use the
Composition Calculator utility.
2.2.1.2.1. Initial (IVC) NOx for High EGR Cases

When high EGR is present and the engine is expected to generate a lot of NOx in the engine-out exhaust,
then it may be necessary to adjust the initial gas composition to include some small fraction of NO in
the initial conditions. This can be important as significant amounts of NO (that is, greater than ~50
ppm) can impact the ignition process through NOx-fuel sensitization effects. In addition, with high EGR,
the presence of NOx in the exhaust can feed back to the overall measured exhaust of NO.
Best Practices
The molecular weight of NO (30 g/mole) is about the same as air or vitiated air (~29 g/mole). For this
reason, adding 25 ppm (0.000025 mole fraction) of NO is approximately equivalent to adding 0.000025
mass fraction. For a more exact calculation, you would need to convert the mass fractions of the initial
composition to mole fractions, add the ppm NOx, re-normalize, then convert back to mass fractions.
2.2.1.3. Fuel Mass for Direct Injection Cases

ANSYS Forte requires the total mass of fuel injected, which is assumed to be per cylinder, and per cycle.
In contrast, often the data available is per engine or per stroke. If data are provided per engine, then
you need to divide by the number of engine cylinders for the input to Forte. If the data is provided per
stroke, then the conversion is as follows:
If the per-stroke fuel data [mg/stroke] is given for one cylinder, then the Forte input is:
Injected fuel mass [mg/cycle/cylinder] = fuel [mg/stroke/cylinder] * <number of strokes(4 or 2)>
If the per-stroke fuel data [mg/stroke] is for the whole engine, that is, for a multi-cylinder engine, then
the input to Forte is:
2.2.2. Basic Calibration Procedures

2.2.2.1. IVC Pressure and Temperature
With IVC pressures and temperatures, there are typically uncertainties in these values. The temperature
value is typically a derived value, based on the measured pressure, mass of air + EGR in the cylinder at
IVC, and the ideal-gas relation. The pressure measured at IVC has higher uncertainty than the pressure
measured after compression has begun, prior to fuel injection. For this reason, the standard procedure
is to do one round of iteration to determine the pressure, as follows:
1. Use the given initial Pand T and run the case to ~2 CA before start of injection.
2. Compare the predicted pressure profile up to this point with the measured values. If the pressure is too
low or too high just before start of injection (SOI), adjust the IVC pressure value. However, this adjustment
should be no more than ~3 PSI, which is less than ~0.2 bar.
3. Since the measured temperature is derived from the pressure and it is important to keep the total mass in
the system consistent with measured values, also scale up the IVC temperature according to the adjustment
made to pressure. In other words, if you change the initial pressure from 1.0 bar to 1.1 bar, also increase
the initial temperature by a factor of 1.1 (that is, if it was 400 K, it should now be 440 K).
2.2.2.2. Injection Timing

Possibly due to inadequacies of the surrogate-fuel model used, adjusting the injection timing as much
as ~1–2 degrees CA to get reasonable timing for ignition is considered acceptable. The total mass of
the fuel into the system, however, is accurately measured, as is the total air in the system, so these
should not be adjusted. The main reason to make this adjustment is for comparison of emissions (in
addition to ignition) with a focus on trends, rather than just seeking how well a particular mechanism
provides ignition. Note that if you do not have the pressure trace close to agreement with experiment,
there is little reason to compare emissions, because they will be at least as wrong as the pressure profile
(which is an indicator of the temperature profile in the simulation).
The best way to judge the injection timing is to compare the heat-release rate between prediction and
experiment. Both are usually derived from the pressure trace, but this comparison shows more clearly
the ignition timing than the pressure trace itself. The initial ignition behavior will be strongly dependent
on the injection timing. The following adjustments may be acceptable:
1. If your heat-release peak is shifted by a couple of crank angle degrees relative to experiment, then you can
shift the SOI of injection accordingly. This assumes that the width of the HRR curve is about the same as
the data.
2. If the predicted HRR curve shows a higher peak and skinnier profile than the data, you might need to adjust
the duration of injection. (Or, conversely, a low peak with an overly fat HRR may need a shorter duration.)
3. If there are multiple injection events, then this problem becomes more complicated. You may be able to
see the HRR signature from the different injections, depending on how much heat is released as a result,
but it may be hard to justify independently moving one peak relative to another or changing fuel splits.
One would expect that the timing delay between actuator and fuel injection would not be different for
different injection events, so the safest bet would usually be to shift all events together.
2.2.3. Advanced Calibration

In general, we don't want to be calibrating beyond the basics. However, it is possible that some of the
standard spray or turbulence parameters might need to be adjusted in the case of the engine conditions
or fuel properties being outside of conventional heavy-duty diesel-engine operation. Only a few such
parameters typically have noticeable impact on the results. The following provide some guidelines as
to the effects to expect and the degree to which a parameter has an impact on the observed engine
behavior. Here the degree of the effect (High, Medium, Low) indicates the effect relative to other
parameters listed. All of these parameters have a secondary, minor effect compared to the basic calib-
ration described above. Note that while the Discharge Coefficient has the largest effect of these, it may
not be considered an “adjustable” parameter, if some attempt has been made to estimate it based on
a nozzle-flow model, for example.
Table 2.1: Parameters providing secondary impacts on engine behavior
Typical observed effect of

increasing this parameter on:
Relative Parameter Recommended Default Physical effect of Peak Expansion Ignition
effect label range value increase pressure pressure timing
(H,M,L)
H Nozzle 0.7–1.0 0.7 Decreases the -- Lower Delayed;
Discharge injection velocity; lower
Coefficient increases initial HRR
droplet size. Effect slope
on droplet size is
typically dominant,
which leads to
slower vaporization
rate, delayed
ignition.
M to RT Distance 1.7–2.5 1.9 Delays secondary -- -- Delayed
H Factor breakup and
therefore
vaporization
Best Practices
M KH Time 20–80 40 Delays spray -- -- Delayed;

constant breakup near lower
nozzle HRR
slope
M Entrainment 0.3–0.7 0.5 Enhances Lower Lower Lower
constant entrainment of air slope of
into fuel jet HRR
M Fuel droplet 350–500 400 Increases enthalpy Increase Increase Earlier;
temperature of fuel jet Increase
(K) HRR
slope
L KH size 1.0 1.0 Delays spray -- -- --
constant breakup near
nozzle
L RT size 0.1–0.2 0.15 Delays breakup -- -- --
constant away from nozzle
L RT time 1.0 1.0 “” -- -- --
constant
L Collision 0.2 0.2 Collision model -- -- --
radius of range
influence
L Cylinder head -- -- Could reduce liquid -- -- --
temperature on head
L Chemistry Defaults 10 K, Resolution of -- -- --
solver: DCC 0.05 clusters
parameters phi
L # Particles 2000–5000 2000 Resolution of -- -- --
injected per droplets
nozzle hole
2.2.4. Comparing to Experimental Data

2.2.4.1. Calculating Apparent Heat Release Rate from Experimental Pressure
ANSYS Forte will calculate apparent heat-release rate automatically from the predicted pressure profile.
This data is stored in the Forte solution file as Net HRR. In some cases, however, the experimental data
provided only include the pressure measurement and not the derived heat-release rate. In other cases
the heat-release rate is provided, but because there are some uncertainties in how that data was derived
from the pressure profile, it may be better to calculate it so that the method is consistent with the
model data and the comparison between simulation and experimental data is consistent.
The HRR profile can be calculated directly from the pressure vs. CA data in Excel, using the following
formulation:
Here Q is the heat-release, and is the heat-release rate, which is usually desired in J/degree units.
CA is the crank-angle, gamma is the specific heat ratio of the gas, which we take to be 1.35, V is the in-
cylinder volume, and p is pressure. Note that Forte uses this same formula in the calculation of the HRR
in the thermo.csv file.
If the volume of the chamber is not available, it can be calculated from the crank angle and engine
parameters. First calculate the position of the piston, which will require the connecting-rod length (L)
and the crank radius (R). Then, use that along with the engine Bore and the clearance volume to calculate
the total volume in the cylinder at each crank angle. The calculations are as follows:
1. Piston position, is defined as follows:
2. Volume can then be calculated as:
3. At the first CA point: dQ=0
4. At CA point n, where n = 1, .., last:
5. Be careful with unit conversions! …It’s a good idea to compare the calculated volume profile with
that calculated by Forte to make sure no mistakes have been made in steps 1 & 2.
Note: You can create a dummy thermo.csv in a folder that resides as a parallel-level directory to
your run directory and put the CA vs. AHRR data into that file for direct comparisons during run-time
using the Monitor plotting. For example, create a directory called xData_Run1 and place it in the
same parent directory of the Run001 directory, then copy any thermo.csv file into the xData_Run1
folder; edit it and replace the CA, P and AHRR columns with the experimental data (with the units ex-
pected in thermo.csv), but zero out all other columns. Now your plot will show your run-time com-
parison in the Monitor window.
2.2.4.2. Importing Data Arrays into Forte Visualizer

Once you have the Pressure and HRR data in Excel, you can easily import it into the ANSYS Forte Visu-
alizer for direct comparisons with predictions. First, create a separate sheet that has just the values of
CA, P, V, and HRR in separate columns. Create column headers that indicate the label and give the units
in Forte-recognizable units-strings. These could be, for example:
1. CA_ATDC (degrees)
2. P_Case1 (bar)
3. Volume (cm3)
4. HRR_Case1 (J/degree)
With the columns of data set up in this way, save the sheet as a comma-separated-values text file (.csv)
using Excel’s “save as … other format” option.
Best Practices
From the ANSYS Forte Visualizer, you can now import the data, using the File > Import menu command.
Allow the software to read the labels from the csv file. It should parse the unit strings in parentheses
and then will know the units context, such that conversions can be made using the Units Preferences
options.
2.2.4.3. Review Spray Behavior

To make sure everything was set up correctly, always review the spray injection and vaporization beha-
vior in the ANSYS Forte Visualizer. Make sure that the injection timing aligns with the SOI/DOI values
provided in the setup, and that the spray is pointed into the piston bowl.
For a “good engine” we’d expect the combustion to be taking place primarily in the bowl and that the
droplets are getting consumed before they hit the wall. If spray seems to be impinging on the wall
(undesirable) this could point to issues with the spray model, such as the discharge coefficient being
too high, for example. If spray droplets are migrating out of the bowl and persisting into the expansion
stroke, then the vaporization or fuel model may not be the best for this case. While it is possible these
types of phenomena are real, it is good to first check whether inputs are reasonable and that the results
are not sensitive to some estimated parameter.
Another good check is to confirm that the total mass of fuel injected, as reported at the end of the
Forte log file, matches the value input in the spray injection data. This should be the total fuel injected
for all nozzles (that is, not just the one nozzle represented in the sector), since that is what is measured
and Forte takes care of the periodicity internally.
2.2.4.4. Emissions and Performance Data

The emissions and engine-performance data can be found in summary form at the end of the
FORTE.log file in the working directory of each run directory. Since these are single-point data, they
are not otherwise accessible from the Visualizer. The Forte log contains the following example of summary
data:
• IMEP (Indicated Mean Effective Pressure)
1. ISFC (Indicated Specific Fuel Consumption)
2. Thermal Efficiency
3. Emissions indices for NOx, CO, UHC (unburned hydrocarbons), and soot, in the units of g/kg-fuel. These
values are at the end of simulation, usually exhaust-valve opening, so they are engine-out conditions,
not exhaust-system-out.
In addition, species emissions in the units of PPM can be determined from the molefraction.csv file
found in the working directory of each parameter-study run. To get EVO emissions numbers, scroll down
to the EVO crank angle (last row) and then go to the column for the molecule of interest. Note that the
fraction of NOx is the sum of the fractions of NO + NO2. PPM is 106 times mole fraction.
Note that, for an in-cylinder-only simulation, the IMEP and ISFC are calculated based on integrating the
P-V curve for the compression and expansion strokes (i.e., -180 CA ATDC to 180 CA ATDC, where 0 CA
ATDC is the firing TDC), which is different from what is measured. Typically, an in-cylinder-only simulation
starts from IVC and ends at EVO. The P-V curve between IVC and EVO is integrated directly. The integ-
rations for “-180 CA ATDC to IVC” and “EVO to 180 CA ATDC” are approximated. For example, the work
for “EVO to 180 CA ATDC” is approximated as 0.5*(PIVC+PEVO)*(VEVO-VBDC), in which PIVC and PEVO are
the pressures at IVC and EVO, respectively; VEVO and VBDC are the volumes at EVO and BDC, respectively.
Also, measurements are usually based on brake power, rather than indicated power. So we expect these
Setting Up Automatic Mesh Generation Cases With Valve Motion
to be approximate rather than an exact match, relative to the data; however, we do expect the predicted
results to follow the correct trends.
2.2.4.4.1. Conversion to Mass per Time Units

In some cases, units of [g/hour] are preferred to the emission indices that are in [g/kg-fuel]. The conver-
sion accounts for the fact that the total fuel injected into the engine system is “per cycle” and a cycle
is two revolutions of the crank. As discussed in SAM Settings for Spray Vaporization (p. 25), the fuel
input to Forte is “per cylinder” and “per cycle.” In this way, knowing the total fuel injected per cycle per
cylinder and knowing the revolutions-per-minute speed of the engine as well as the number of engine
cylinders, we can obtain the [g/h] units as follows:
Emissions [g/h] = Emissions [g/kg-fuel]
* Total_fuel_injected [kg/cylinder/cycle]
* (RPM/Revolution_per_cycle) * 60 [cycle/h]
* <number of cylinders> [cylinder/engine]
where Revolution_per_cycle is 2 for a four-stroke engine and 1 for a two-stroke engine.
2.3. Setting Up Automatic Mesh Generation Cases With Valve Motion

You must pay attention to several considerations when working with a geometry that includes valve
motion.
General notes on the use of ANSYS Forte and configuring it for your use:
• The geometry units must be cm. If the surface mesh does not use cm units, please choose the correct unit
in the Import dialog and Forte will automatically scale the geometry to use cm units. If you forget to specify
the correct unit, simply select all the surfaces in the Geometry tree, right click, select Transform and specify
the correct scaling factor for the X, Y, and Z direction.
• Units are completely configurable in Forte, both in the Setup and the Visualizer. Units can be modified inline
in the input panels, or by going to Edit > Preferences > Units Preferences. You can chose from CGS, SI,
British, or create your own Custom unit preferences.
• Have the surface mesh file written from the CAD program so that the surfaces are different parts inside the
surface mesh file, or so each unique boundary is a separate surface mesh file. This saves time since you will
not need to split the geometry.
• The geometry should be watertight in Forte. If the surface mesh is not watertight, Forte will identify the region
where the issue occurs in red.
• The surface mesh file should have the piston at TDC.
• The valves should be closed with a very small gap, but not coincident.
• Specify approximately 10 GB of memory for Forte Simulate (if generating preview meshes) and for the
Visualizer. Use the Launcher Preferences in the Forte Launcher.
Best Practices
2.3.1. Read In and Process the Surface Mesh

2.3.1.1. Selecting the Source of Surface Meshes
The best surface mesh data include surface topology information. The Fluent mesh format and CGNS
format include topological information about the surface and are good choices for import.
Whenever possible, STL files should be a last choice as a source of surface mesh data. STL files lack
surface topology information. As different programs may infer different topologies from the same source
file, there is no way to reliably confirm or test whether an STL is, for example, watertight.
Currently, the best alternatives to STL files supported by ANSYS Forte are Fluent mesh format and CGNS
format. Both of these formats encode topological information about the surface.
A recommended procedure for generating a surface mesh for importing to Forte is:
1. Create a “Mesh” object in ANSYS Workbench and import your geometry, then launch the mesher.
2. Configure the mesher to CFD, Fluent, and specify these options:
a. Defaults:
Physics Preference = CFD
Solver Preference = Fluent
b. Sizing:
Transition = Slow
c. Defeaturing:
Automatic Mesh Based Defeaturing = ON
Other settings may also be adjusted for good results.
3. Generate just the surface mesh by right-clicking the Mesh object > Preview > Surface mesh.
4. Go to File > Export and save the .mesh file.
5. In ANSYS Forte, import the surface mesh using the From Fluent mesh import option; select meters as the
units.
6. When the import gets to the “Checking surface” phase, cancel it with the red X. (Otherwise the surface
checker will report problems because the normals are reversed; see the next step.)
7. Reverse the normals for the imported surface—everything should go from dull to shiny.
8. Run the surface checker manually (from the Geometry toolbar). It should now pass.
2.3.1.2. Splitting the Geometry

If the geometry comes into Forte as one complete surface, you will first need to split it into major
components. This can usually be accomplished with a simple application of the Split utility in the Geo-
metry Workflow tree, using the default feature angle.
• If you end up with some small pieces, try to collect those and merge them into the bigger pieces, using the
Merge utility.
• If you need to split surfaces further, consider using the cut-plane option or adjusting the feature angle that
was used for the Split utility.
Ultimately, you want to end up with the following surfaces that are separated and free of holes:
1. Individual valves
2. Exhaust port wall
3. Intake port wall
4. Cylinder head
5. Cylinder liner
6. Piston surface (usually including bowl)
7. Inlet boundary
8. Outlet boundary
It is easiest if the separate boundaries, such as ports, valves, etc., are written as separate parts in the
surface mesh file or as separate surface mesh files all together. You can select multiple surface mesh
files to read into ANSYS Forte. (The surface mesh must be watertight.) Once again, the geometry should
be in units of cm in Forte. If the geometry is not in cm, specify the unit used for the geometry and Forte
will convert the geometry to cm units upon reading it into the user interface.
Figure 2.2: STL dialog when reading STL files into Forte
The valve stems should be cut and flush at the location where they meet the port (at the end of the
valve stem/port connection).
Best Practices
Figure 2.3: Green region is the valve, pink is the port. Note the donut surface around the
intersection of the two
Note
If the crevice volume is included in the geometry, this volume should be resolved sufficiently
using surface refinement such that there is at least one layer of cells between the two walls
of the crevice gap. However, resolving the crevice gap using tiny cells will increase the cell
count of the mesh and potentially affect solution speed. Therefore, it is desirable to not include
the crevice region in the geometry unless it is absolutely necessary.
2.3.1.3. Normals
In the surface mesh geometry, the normals all need to be consistent. The normals should all point to
the solid region, so for the valves, the normal should point to the center of the valve.
If the normals for some triangles are not consistent, the geometry is not considered to be watertight
and these triangles will be highlighted with a red outline.
• To separate these inverted triangles, go to the Workflow tree and select the surface that contains the inverted
triangles. Split Mesh and select the Feature Angle option. On the Feature Angle dialog, specify a Feature
Angle of 175 as shown in the figure below.
• This will split off all inverted triangles, and you can then use the Reverse Normal tool to correct the
normal.
• Note that the geometry will not be watertight at this point since the triangles are not considered to share
vertices. To fix this, the final step is to use the Join Meshes (not to be confused with the Mesh Merge) tool
to join all the surfaces back together. At this point, the red outline of the offending triangles should disappear.
Note
If you get a mesh-generation failure during preview, an incorrect surface normal is a

very likely cause.
2.3.2. Set Up Mesh Size and Mesh Refinement

2.3.2.1. Material Point
The Material Point marks the location where Forte will generate the mesh, that is, marks what is con-
sidered inside and outside of the geometry where the surfaces mark the boundaries of the domain.
Select a point that will be inside the cylinder at all times during the cycle. Usually this means just
below the head, at the cylinder axis, above the highest point of the piston bowl at TDC. Be sure the
material point is not placed in-line with any of the valves.
2.3.2.2. Global Mesh Size

The Global Mesh Size provides the overall background mesh size. The Global Mesh Size is used every-
where a refinement is not specified. In ANSYS Forte, a suggested global mesh size of 2 – 2.5 mm is
suggested. Mesh size and simulation turn-around time requirements are often competing forces, thus
one must balance these two.
A Small Feature Deactivation Factor can be used to remove very tiny features from consideration during
the computation. This factor is applied globally to the geometry. The value ranges from 0 to 1, where
a higher value will filter out more small features. The default is 0.25 and it is suggested that this option
be left at the default value.
A recommended global mesh size is 2 mm. But for large-bore diesel engines, it may be acceptable
to use 3 mm for the flow portion of the cycle, while for a smaller-bore gasoline engine 1.5 mm may be
more appropriate. You can then use region-based refinement to set a smaller mesh size (for example,
1/2) in-cylinder during spray/combustion.
2.3.2.3. Mesh Refinement

ANSYS Forte uses automatic meshing using Cartesian cells to mesh the geometry. The actual meshing
is done automatically on the fly by the Forte solver, however, the user needs to specify refinements to
the mesh to resolve the shape and features of the geometry.
Mesh refinements are used to resolve both surfaces and regions within the computational geometry.
Note that refinements should be added to sufficiently resolve all crevices in the geometry, that is, small
physical gaps in the geometry, otherwise robustness may be affected. In all cases, refinements are added
in 1/2!, which would be 1/2, 1/4, 1/8, and 1/16.
Each of the mesh-size controls can be specified as static (Always active) or dynamic (active only During
specified time or crank-angle intervals). For example, refinement in the squish region near a moving
piston boundary may only be necessary when the piston is near top-dead-center (TDC) position. For
Best Practices
simulation of a full cycle, this may mean setting two controls for the two intervals (for example, 350 to
370 CA, 710 to 730 CA) during the cycle of a 4-stroke engine while the squish region is most active.
ANSYS Forte has several refinement types that are useful for resolving the geometry. They fall in two
main categories: (1) surface refinement, and (2) volume refinement. Volume refinement is further divided
into two categories: (2a) fixed volume refinement, and (2b) Solution Adaptive Meshing (SAM). Surface
refinement should be used in every case. SAM with limited fixed volume refinements is recommended
for volumes.
1. Surface Refinements:
Applied to surfaces to refine the shape. Certain areas of a surface, such as complex geometries, open
boundaries, and valves, require additional attention to accurately resolve the geometry and moving
surfaces.
Figure 2.4: Suggested refinements for typical engine cases (p. 16) shows the suggested refinements
to be used for most engine cases. Details of each refinement are given.
Figure 2.4: Suggested refinements for typical engine cases
• all_walls: Surface refinement applied to all wall boundaries as a base setting. Typically, ½ refine-
ment is used (Active = Always).
• open_boundaries: Surface refinement applied to the inlets and outlets of the domain. Typically,
½ refinement is used and the number of layers to extend is specified as 2 (Active = Always).
• squish_ftdc and squish_ltdc : Surface refinement applied to the head, piston, and liner surfaces
for the TDC events. Typically, ¼ refinement must be added such that the TDC clearance height is
resolved (Active = During).
• valves: Surface refinement applied to the intake and exhaust valves. Typically, ¼ refinement is
required on the valve. The valve may be split into two parts, stem and face, to save a mesh count
with deeper refinement applied only to the face near the seating area. If more refinement is desired
near the valve seating area, ¼ refinement can be applied to the head or port regions.
• crevice: If a crevice volume is included between the piston edge and the liner wall, this area
should be resolved sufficiently using surface refinements such that there are 2 cells across the
crevice gap. Typically, 1/8 refinement is needed with multiple layers to fill the crevice gap. To reduce
the mesh count, the piston or liner surface can be split to apply deeper refinement only to surfaces
that form the crevice (Active = Always). (See Modeling the Crevice (p. 20)).
• spark_plug: Point refinement applied to the spark plug. Typically, 1/8 refinement is used to suffi-
ciently resolve the geometry and the spark gap, but this will depend on the spark plug geometry.
The point refinement radius is typically specified such that 6 to 10 layers will resolve the region
around the spark plug and the resulting flame front. The refinement should result in about 3 cells
across the spark gap (Active = Always).
• Injector and glow plug: Surface refinement applied to the injector tip. Typically, ¼ refinement
is used to sufficiently resolve the geometry and the number of layers to extend refinement is
typically 2 (Active = Always).
• Feature refinement: This type of refinement can be specified on single or multiple surfaces. A
feature angle is specified to identify features on surfaces (smaller angle = greater feature refine-
ment). The user also specifies a radius that is applied to each identified location. This option is
useful when refining oddly shaped geometry.
2. Volume Refinements:
Volume refinement is important to achieve mesh-independent results. SAM is recommended as it

does not require prior knowledge of the solution field and it saves time and mesh count while
producing accurate results. In addition to SAM, some fixed-volume refinement is recommended for
certain cases.
(2a) SAM controls:
SAM should be used to resolve the flow field and reactive zone. Recommended options are using
velocity gradients and temperature gradients with statistical sigma threshold of 0.5, and ¼ refinement.
Velocity control should be active always and be applied to the entire domain (Figure 2.5: SAM control
for resolving flow fields (p. 18)). Temperature control can be limited to the cylinder region and be
activated prior to TDC for the combustion duration, as shown in Figure 2.6: SAM control for resolving
reactive regions (p. 19).
Best Practices
Figure 2.5: SAM control for resolving flow fields
Figure 2.6: SAM control for resolving reactive regions
Additional SAM controls should be used for accurately capturing the flame front and spray vaporiz-
ation. These are described in Models (p. 21). Additional SAM controls on species such as CO or soot
precursors (like acetylene and pyrene) may be used to better resolve emissions but they are not
necessary.
(2b) Fixed volume refinements:
Fixed volume refinement should be used to provide some level of initial refinement in the regions
involving combustion and spray. Following these fixed refinement recommendations should provide
a good base mesh and help to achieve high fidelity results when used together with SAM.
• Secondary Region:
Under the Geometry node, secondary regions can be defined by selecting geometry to encapsulate
a region. Under the Mesh Controls, the secondary regions can be used to refine an entire region
(in other words, the chamber). Typically, this is a secondary volume encapsulating the in-cylinder
Best Practices
volume that can be created by selecting the surface that includes liner, piston, and head. The re-
finement level for the in-cylinder region should be set to ½ or ¼ to achieve a cell size no larger
than 1 mm and set to Always Active.
• Line Refinement:
The end points of a line and a radius are specified to define a cylinder where all locations inside
the cylinder are refined to the specified refinement level. Line refinement of ½ to ¼ should be
used to resolve near-nozzle regions during spray duration. Point refinement may be used instead
of line refinement depending on the geometry.
2.3.2.4. Modeling the Crevice

While it is recommended against modeling long narrow crevices (unless there is some engineering
reason to do so) due to the computational overhead, such crevices may be included if necessary. The
best approach to model the crevice is to do the following:
1. Have a separate surface for the bottom of the crevice
2. Split the piston so the side of the piston becomes a separate surface
3. Apply 1/8 or 1/16 refinement to the crevice bottom surface and the piston-side surface. The crevice thickness
will determine the refinement level to be used, but 1/8 or 1/16 is typically used to resolve the thickness.
4. If possible, modify the geometry such that the crevice thickness can be greater while capturing the same
volume, to avoid many highly refined cells.
2.3.2.5. Modeling Sector Geometry Using Automatic Mesh Generation

Forte allows cylindrical sector geometry (with periodic boundary conditions) to be meshed using
automatic mesh generation. This is complementary to the body-fitted sector mesh feature described
in Sector Mesh Generation (p. 3). Since the automatic mesh generation in Forte uses the Cartesian
mesh with immersed boundary method, it is necessary to make sure that the background Cartesian
mesh cells near the geometry boundaries are sufficiently refined, especially near the axis of the sector.
To avoid display artifacts in post-processing, we recommend that a line refinement be applied along
the sector axis. The following are recommended settings for this line refinement:
• End Points: This line refinement should cover the full length of the axis when the piston is at the BDC position.
The end points can be set to (X=0, Y=0, Z1) and (X=0, Y=0, Z2), in which Z1 is the Z-coordinate of the top
end of the axis on the cylinder head, and Z2 equals the Z-coordinate of the piston center at the TDC position
minus engine stroke (assuming the piston moves towards the negative Z direction from TDC to BDC).
• Radius of Application: The radius should be larger than half of the global mesh size. Recommended practice
is to set the radius equal to the global mesh size.
• Size as Fraction of Global Size: The cell size of this refinement should be equal or smaller than the cell
sizes from other refinements. The goal is to maintain uniform cell size within the range of this line refinement.
• Active: This refinement should be set to be Always active.
2.3.2.6. Multi-Cylinder Simulations
Note
Best practices for multi-cylinder simulations are described by the tutorial in the Forte Tutorials:
Multi-Cylinder Four-Stroke Engine Simulation.
2.3.3. Models
2.3.3.1. Turbulence Model
In modeling turbulent flows in internal combustion engines, the Reynolds-Averaged-Navier-Stokes
(RANS) approach is most efficient. It simulates the ensemble average of turbulent flow and transport
processes. The default option, RNG k-ε model, is recommended for most of the engine flow and com-
bustion problems. In general, there is no need to adjust the default model parameters.
The Large-eddy simulation (LES) approach is intended to simulate individual flow realizations rather
than the ensemble average. It is useful when simulating cycle-to-cycle variations of flows and combustion
outcomes in engines. The LES Smagorinsky model with its default model parameters is recommended,
due to its numerical stability in cases where complex engine geometry is considered. When the LES
approach is used, we suggest turning on the option of using central difference for momentum convection,
due to its better resolution in solving the large-scale momentum transport.
In general, LES requires finer mesh resolution than RANS to better resolve the turbulent transport pro-
cesses.
2.3.3.2. Flame Speed Model for SI Engine Simulations

The Flame Speed Model under the Chemistry node can be used to control the flame propagation process
for a spark-ignited engine simulation. Using the Table Library option for the flame speed model is re-
commended.
Typical values for the parameters that control flame speeds are shown in Figure 5 (p. 22).
Best Practices
Figure 2.7: Turbulent flame-speed input parameters
The turbulent flame speed is a function of the laminar flame speed. The Table Library model contains
Chemkin-based laminar flame speeds and computes flame speeds for the multi-component fuel on the
fly. It captures the influence of fuel composition, pressure, temperature, AFR, and EGR on the laminar
flame speed. A Flame Library should be created that includes fuel components when premixed or direct-
injection fuel surrogate are used in the chemistry. An example flame library for a TRF gasoline are shown
in Figure 2.8: Example of a Flame Library for a TRF gasoline (p. 23).
Figure 2.8: Example of a Flame Library for a TRF gasoline
The Turbulent Flame-Speed Ratio (b1) and the Stretch Factor Coefficient are the two inputs that may
require adjustment to match the pressure curve. See the tooltips and the Forte User's Guide for more
details on these inputs. Since the model formulation for the stretch factor is very sensitive to its input
parameters, including laminar flame speed, laminar flame thickness, and turbulence parameters, the
stretch factor may show strong fluctuation under strong turbulence and low laminar flame speed con-
ditions. Thus, it is recommended to keep the default setting of 0 for the stretch factor coefficient, and
adjust only b1. In the turbulent flame-speed correlation used by Forte, b1 is closed using experimental
data and thus can be adjusted for model calibration. A larger b1 will result in higher turbulent flame
speeds and faster flame propagation, and it has an impact throughout the flame propagation process.
If unable to match the pressure curve during flame propagation by adjusting b1, the Flame Development
Coefficient (FDC), which is another parameter available on the Spark Ignition Editor panel, can be adjusted
to improve flame propagation prediction. A larger FDC can result in faster growth of the flame from
ignition kernel flame stage to fully-developed turbulent flame, which in turn leads to faster burn rate.
FDC typically has larger impact on the early stage of flame propagation than the later stage. For the
other settings, default values are appropriate for most cases.
Best Practices
2.3.3.2.1. SAM Settings for Flame Propagation

For achieving accurate and mesh-insensitive flame propagation, we recommend using SAM (solution
adaptive mesh) controls with the absolute range solution variable “G” as shown in Figure 2.9: SAM settings
for flame propagation (p. 24). This setting will help refine the region around the flame front as defined
by G=0. This refinement should be activated typically at spark ignition and typically for 60 degrees
longer until combustion is complete. The recommended cell size for accurately capturing the flame
front is no larger than 0.6 mm.
In the G-equation model, the non-reacting scalar G is defined at each mesh vertex for the purpose of
marking the mean flame front location. The G value at a vertex is tracked as the signed normal distance
from the vertex to the nearest flame front surface. G < 0 in the unburned zone, G > 0 in the burned
zone, and the G=0 iso-surface marks the flame front location. Since Forte uses the CGS unit system, the
default unit for G is cm, although G is handled as a dimensionless quantity in the code and it involves
no unit conversion. In the example shown in Figure 2.9: SAM settings for flame propagation (p. 24), the
range of [−0.5, 0.25] means that the SAM refinement is applied within a band surrounding the flame
front. The bandwidths on the unburned side and burned side are 0.5 cm and 0.25 cm, respectively.
Figure 2.9: SAM settings for flame propagation
In addition, it may be productive to use additional SAM controls with gradients of either/both temper-
ature and of CO. Using temperature/CO control can be helpful for high compression conditions where
there could be a possibility of autoignition or reactivity ahead of the flame front. We suggest statistical
bounds with a Sigma Threshold of 0.5 to 0.75 for temperature. Using a smaller sigma value results in
more refinement. This refinement should be activated at or slightly prior to spark timing.
2.3.3.3. Spray Model

In general, most of the defaults can be used for the spray model for the injector. The default Collision
Mesh model for droplet collision is suggested for better computational speed. For solid cone injectors,
a Droplet Density option with a value of 1 is recommended. When nozzle dimensions are available,
using the Nozzle Flow model may provide better spray initialization at the nozzle exit. However, for
gasoline sprays using Rosin-Rammler or Log-Normal droplet size distribution, explicitly specifying the
discharge coefficient should provide better representation of the injector exit. The discharge coefficient
for multi-bore high pressure injectors can be as low as 0.5, which is lower than that for most diesel in-
jectors.
For diesel fuel, specifying Uniform Size for the droplet size distribution is suggested. For gasoline sprays,
Rosin-Rammler distribution is more commonly used than log normal distribution. The Initial Sauter
Mean Diameter should be specified. For most gasoline cases, this is typically between 10–30 micron.
The actual droplet distribution is subject to an upper limit, which is the (nozzle diameter * Cd0.5).
For solid cone breakup, time constants for the KH model (primary breakup) can be adjusted within the
range suggested in the tool tip. The check box to Impose SMR Conservation in KH breakup should
be left unchecked for most conditions. However, it can be turned on for certain conditions such as low-
load low-temperature combustion diesel engines, for improving the pressure curve and emissions pre-
dictions.
The RT model for secondary breakup is important for most conditions. The RT distance constant can
be adjusted to improve the prediction of both spray penetration and combustion. A lower value of the
RT distance constant, such as 0.2, will activate the secondary breakup model closer to the nozzle. This
may be important for gasoline atomizers where spray breakup occurs very close to nozzle. For most
diesel engines, the RT distance constant typically varies within 40% of the suggested default value of
1.9. Refer to Table 1 (p. 7) for the typical qualitative impact on engine behavior.
When creating nozzles, first create one nozzle and specify all its settings, and then Copy it and Paste
repeatedly to create the remaining nozzles, modifying the copied settings as needed. This saves time
and only one parameter needs to be adjusted to change the orientation. The Copy and Paste buttons
are located at the top of appropriate Editor panels.
Figure 2.10: Copy and Paste buttons located at the top of Editor panels
2.3.3.3.1. SAM Settings for Spray Vaporization

Spray vaporization can be quite sensitive to cell size. Therefore, it is advisable to use finer mesh to ac-
curately account for vapor distribution and mixing. As shown in Figure 2.11: SAM settings for spray va-
porization (p. 26), SAM controls should use the gradient of solution field on variable FuelVaporMassFrac-
tion with statistical sigma threshold between 0.5 to 0.75. Smaller sigma value result in more refinement.
Recommended cell size for accurately capturing the vaporization is no larger than 0.5 mm.
Best Practices
Figure 2.11: SAM settings for spray vaporization
2.3.3.4. Spark Model

For the spark model, the duration of the spark is typically about 10 CA degrees, the energy release rate
is about 20 J/sec, the energy transfer efficiency is 0.5, and the initial kernel radius is about 0.5 mm. The
initial kernel radius should be sufficiently small to fit in the spark gap, but remember that the spark gap
should also be sufficiently mesh-refined.
2.3.3.5. Soot Model

Forte has three options for modelling soot:
1. The 2-step soot model: The Soot model in Forte where only acetylene (C2H2) is the precursor to soot
formation. Several values are available for adjusting the soot formation and oxidation. This option should
be selected if acetylene is the only precursor for soot. Note that the gas-phase chemistry mechanism must
contain soot as a species when this soot model option is used.
2. Pseudo-gas soot model: The pseudo-gas soot model was developed by ANSYS and is included with the
gas-phase chemistry set. A gasoline and diesel surrogate and mechanism are provided with the Forte in-
stallation. If you are using the pseudo–gas-phase soot model, the Soot Model option in Forte should not
be selected since all soot-related chemistry is included in the gas-phase chemistry file.
3. Method of Moments: The most detailed option for modelling soot is particle tracking. ANSYS Forte will
predict the average particle size and number density for every cell in the computation. To use this option,
choose the Method of Moments. There are two mechanisms for particle tracking included with Forte, one
for gasoline and one for diesel fuel. The Method of Moments option requires specifying the soot surface
chemistry mechanism.
2.3.4. Boundary Conditions

2.3.4.1. Boundary Profiles
In ANSYS Forte, it is very easy to use profiles for time-varying boundary conditions, for example, valve
lifts. When specifying time-varying profiles for engine cases in Forte, the profile should go from 0 to
720 CA for a four-stroke engine and from 0 to 360 CA for a two-stroke engine. By specifying the profile
in this manner, it is easy to then run multi-cycle engine simulations. Profiles can be copied from Excel
and pasted in the Profile Editor.
2.3.4.2. Valves
In this section, the setup of the intake valve is described. The same setup and procedure would be used
for the exhaust valves. We suggest that you set up one valve, then Copy and Paste it to create the
other valves.
In this setup, the valve stem has been split off from the valve face, as shown in the figure below. This
has been done to improve the speed in the calculation where Forte determines if the valve is open or
closed, that is, the specified gap between the valve and valve seat has been obtained. Splitting off the
valve face from the stem is not required, but it may improve simulation time since the refinement for
the valve opening gap is only applied to the valve face and not the entire valve.
Figure 2.12: Valve stem split off from the valve face
2.3.4.3. Valve Angles

The easiest way to specify valve tilt angles (orientation) is to use spherical coordinate systems. This will
be explained using one example shown in Figure 2.13: Valve tilt angles (p. 28) . In this example, the
engine geometry is oriented such that the engine valves are tilted in the X-Z plane. When looking from
the negative Y direction, the intake valves are on the right and the exhaust valves are on the left. In
this configuration, the spherical coordinate system for each valve requires two angular inputs, θ and φ,
where θ is the valve angle to be specified and φ can be left as zero. The valve angles will be:
Best Practices
Figure 2.13: Valve tilt angles
• Intake Valve Angle(θ) = 180 + valve_angle, for example, if the valves are canted 13 degrees, then the resulting
angle would be 193 degrees.
• Exhaust Valve Angle(θ) = 180−valve_angle, for example, if the valves are canted 12.6 degrees, then the res-
ulting angle would be 167.4 degrees.
2.3.4.4. Valve Lift

For each valve, copy and paste its lift profile into the profile editor. Make sure the profile goes from 0
to 720 degrees and also be sure the units for the profile are correct.
2.3.4.5. Movement Type

In boundary conditions where wall motion is activated, the user has three options for the movement
sub-type:
• Moving Surface: Used for the piston
• Valve: As the name indicates, this is used for valves
• Sliding Interface: Used in ported engine cases where the piston covers/uncovers port openings to the cyl-
inder
For the Valve movement sub-type, the user needs to specify the valve seat and the surface it contacts.
The valve seat is the stationary portion that may be a part of the head or intake/exhaust port surface
and the surface it contacts is the valve itself.
In this case, some portions of the intake port (similarly for the exhaust port) have been split off and
labeled as the “seat”. This splitting is not a requirement, but has been done in this case to help with
the calculation speed of the valve opening/closing events (see Notes in the Valves section above). The
splitting is accomplished with a cut plane in the Geometry tree.
Select both the “seat” portion and the “valve” from the list by holding down the Ctrl key and clicking
with the left mouse to multi-select.
Now that the Valve movement subtype surfaces have been selected, you need to specify the Valve
Motion Activation Threshold, that is, the minimum gap before the valve is considered open or closed.
It is suggested to start with a 0.5 mm gap in the first simulation, and then you can use a smaller gap
if you are not satisfied with the results. The other required input is the Approximate Cells in the Gap
at Minimum Lift and this should be 2. Note that Forte automatically refines the valve gap such that
the specified number of cell layers can fit in the gap, so no special refinements are required for resolving
the gap.
2.3.5. Initial Conditions

In engine simulations, there are unique regions that need to be initialized with different conditions,that
is, the intake, exhaust, and the cylinder. In Forte, there is always a “Default Initialization” which will apply
to the entire geometry unless other initial conditions are created to “override” the default condition.
To add a new Initial Condition, right-click Initial Conditions and select Add > Secondary Region. You
will need to add a total of two secondary regions, one for the intake, and one for the exhaust. The default
initialization will be applied to the cylinder, which is the region not marked by either the intake or the
exhaust secondary regions.
In the Secondary Regions, a location needs to be specified to mark the different regions. This is like a
Material Point for each unique region, so in this case there would be a marker point for the intake and
exhaust regions. The location of these marker points does not need to be exact, but they must be com-
pletely inside the port (region). The figure below shows where to input the location information for the
marker points in the initial condition panel of a secondary region.
Best Practices
Figure 2.14: Indicating the marker point location for a region
2.3.5.1. Initialization Order

In Forte, an initialization order is needed to help define region ID values on cells and vertices at the
interfaces of different regions, especially when the regions are connected by flow. As a rule of thumb,
the orders should be specified in the direction of flow, so they would be:
• Intake: Initialization order = 1
• Default Initialization: Initialization order = 2
• Exhaust: Initialization order = 3
If you are running a sector mesh, or a combustion/expansion stroke only and the valves are not included,
you will only have one initialization (the default) and the initialization order for that region would equal
1.
2.3.5.2. Simulation Controls

• Time Step
• ANSYS Forte uses an adaptive time stepping approach where the user specifies a maximum time-step size.
The maximum time-step size in Forte can be a constant value or a profile, that is, a function of time or crank
angle. The following are suggestions for maximum time-step size:
– Sector Mesh Cases: 8E-06s during compression, 5E-06s during injection, combustion and expansion (Note,
it may be possible to increase the time-step size back to 8E-06s late in the expansion stroke)
– Automatic Mesh Generation Case: Prior to injection or combustion, a maximum time-step size of 1E-05s
can be used, during injection and combustion a maximum of time-step of 5E-06s should be used.
• Chemistry
• There are several options for reducing the simulation time for chemistry or modulating chemical heat release
rate by considering turbulent mixing effects. Below are suggestions for these options:
– Turbulent Kinetics Interaction (TKI): Sub-cell scale turbulence chemistry interaction can be accounted for
with the TKI model. This mixing time-scale model considers that the combustion chemistry should be
partly controlled by the breakup of turbulent eddies due to the imperfect mixing of fuel and oxidizer in
an actual engine process. The TKI model is recommended for use with high-load cases only; for low-load
cases the impact of the TKI model is expected to be much less. Also, the TKI model is recommended only
for cases with in-cylinder fuel injection, for which the TKI model would consider the effects of imperfect
mixing of fuel and oxidizer.
A value of 1.5 is recommended for the mixing time coefficient setting under the TKI model. While
a larger value of this input increases the turbulent mixing effect on reaction rates, a very high value
would also suppress chemistry for post-injection expansion stroke conditions.
Figure 2.15: Turbulent Kinetics Interaction Model (TKI) panel
– Dynamic Adaptive Chemistry (DAC): This option performs a mechanism reduction on the fly during the
simulation. It should only be considered when there are 500+ species in the reaction mechanism. For the
initial species, all fuel species, important radicals (such as OH, HO2), NO, NO2, and soot precursor species
(C2H2, soot). In the example below, the fuel n-heptane (nc7h16), and the soot precursors acetylene (c2h2)
and benzene (c6h6) are included.
For the search tolerance, the default value of 0.0001 is recommended. The Size threshold for ac-
tivating DAC acts as a trigger for activating DAC. This value represents the ratio of species in the
reduced mechanism to the number of species in the full mechanism. DAC will be enabled when
this ratio is smaller than the threshold set here. A value of 1.0 will result in DAC being always en-
abled. Depending on the size of the full mechanism, this value can be set between 0 and 1.
– Dynamic Cell Clustering (DCC): This option lumps together kinetically similar cells, thus reducing the size
of the chemistry problem to solve. This option should always be activated and the settings kept at the
default.
– Activating Chemistry: By default, chemistry is always solved, but this is not necessarily needed prior to
injection or spark timing. Use the Conditional option and activate the chemistry at the combustion event,
that is, injection or spark, and use the Threshold Temperature option.
Note
The chemistry activation controls using spark timing or injection timing will be ignored
after the first engine cycle. Because of this, we recommend using controls based on
crank-angle intervals for multicycle simulations. In a crank-angle-interval–based control,
both the user-specified starting and ending crank angles will be converted to fit in the
range of [0, 720) or [0, 360) CA degrees for four-stroke or two-stroke engines, respectively.
The chemistry activation/deactivation events will consequently be examined in those
engine cycle ranges.
Best Practices
Figure 2.16: Dynamic Adaptive Chemistry (DAC) panel
• Transport Terms
• Typically, these never need to be adjusted. If you are having convergence issues, please contact the ANSYS
support team at support@ansys.com.
2.3.5.3. Output Controls

Spatially Resolved Data
Spatially resolved data is written to every cell during the calculation. While the user specifies the fre-
quency, keep in mind that the amount of disk space required will go up with a higher frequency. Typ-
ically, solutions are written every 10 CA degrees and the User Defined outputs option is used to write
data out more frequently during important points in the cycle, such as injection, spark, combustion.
For File Size Control, there are three options:
• – One File: All the data is written to a single file. This is not the desired option since the file will become
very large and this requires more time during file I/O and in rendering in the post-processor.
– File Size: This option allows you to specify a file size in megabytes for the maximum file size before another
file is written.
– Solution Count: This is the best option since you can control how many solutions are written to a single
file. A value of 1 means one solution per file. This is the preferred option.
• Spatially Averaged Data
• Spatially averaged data is data that is averaged over the domain. Since this is X-Y data, the data can be
written more frequently, such as every 1 CA degree, or even more frequently.
• Restart Data
• Restart points can be written at certain frequencies in Forte. It is suggested a restart be written on the last
simulation time-step, and also during the simulation prior to events occurring, such as IVO, injection, spark.
You specify the times to write out restart points.
2.3.5.4. Preview Simulation

• Prior to submitting the simulation to the Forte solver, you should preview the motion of all the boundaries
to be sure all the moving parts are moving correctly. In addition, it is suggested to preview the mesh before
submitting the run. This will provide some information on the number of cells used and allow you to visualize
the mesh to be sure it is not to coarse and not too fine. Some points of interest are: TDC, injection timing,
spark timing, IVO, peak intake valve lift, EVO.
2.3.5.5. Forte Engine Simulation Checklist

During an IC engine simulation, there are certain points and items to check to ensure the best possible
results are achieved. The following checklist can be used as a guideline:
1. Check CR compression ratio by running no-hydro case from -180 to 0 CA:
Three options are available to adjust compression ratio:
a. Clearance (squish) height
b. Crevice width (for sector case, use between 1.0 and 1.5 mm, avoid a long skinny crevice)
c. Crevice height (can be adjusted for sector case)
2. Run simulation to SOI or spark timing, check compression pressure:
• Adjust intake pressure and temperature to match pressure at SOI or spark timing (scale both P and T by
same %)
3. Check A/F ratio in the simulation vs. experimental:
a. Check IVC mass to be sure A/F ratio computed mass is consistent with P & T for the IVC condition for
sector case.
b. IVC total mass can be computed from the ideal gas equation.
c. If needed, adjust P or T up or down to match IVC mass.
4. Check ignition timing by looking at heat release rate (HRR):
a. Be sure you compare apparent HRR and NOT chemical HRR with engine data; apparent HRR takes heat
losses (computed from P curve).
Best Practices
b. You can adjust SOI +/−1CA deg. (typical uncertainty) to match the ignition timing.
5. Check expansion pressure:
• If expansion pressure is low, there could be several possible reasons:
a. Fuel amount is incorrect.
b. Too much heat transfer through the walls, typically the liner.
6. Check emissions:
a. NOx is very important; if NOx is not predicted accurately, there is little confidence in other emissions,
such as soot.
b. NOx is highly dependent on temperature and can be dominated by local temperatures.
2.4. Setting Up Two-stroke Engine Cases

This section describes the recommended settings for a two-stroke engine simulation. Information from
the section, Setting Up Automatic Mesh Generation Cases With Valve Motion (p. 11), is also relevant.
2.4.1. Piston/Cylinder Wall Gap

Two-stroke engine geometries require some extra care to set up correctly. The extra complexity arises
from the difference in the method used to isolate in-cylinder cells from other portions of the fluid domain.
In a valved engine, valve motion explicitly triggers fluid regions in ports to connect or disconnect from
the in-cylinder regions. In a typical two-stroke configuration, ports are connected or disconnected from
the in-cylinder regions as the piston slides past openings in the cylinder wall. To account for this beha-
vior, the geometry must provide a very tight but non-zero gap between the piston skirt and the cylinder
wall. In other words, the piston must fit just inside of the cylinder wall without any edge intersecting
the cylinder wall.
• If the gap is too large, columns of cells will be created in the crevice, potentially connecting the ports
to the cylinder prematurely. See Figure 2.18: Gap too large (p. 35).
• If the gap is so small as to allow piston edges to bleed through the cylinder wall, the surface topology
becomes invalid and automatic mesh generation will fail, causing the run to abort. See Figure 2.19: Gap
too small (p. 36).
To avoid these issues, the rim of the piston should be resolved at a similar angular resolution as the
cylinder wall, and the gap between the piston and cylinder wall should be no more than 0.5% the cyl-
inder bore (ideally < 0.25%). See Figure 2.21: Mismatch in angular resolution causing self-intersec-
tions (p. 37).
Setting Up Two-stroke Engine Cases
Figure 2.17: Sample geometry with gap demonstration
Figure 2.18: Gap too large
Best Practices
Figure 2.19: Gap too small
Figure 2.20: Acceptable gap
Figure 2.21: Mismatch in angular resolution causing self-intersections
2.4.2. Locating the Piston and Defining Motion

Just as with four-stroke valved engines, Forte requires the piston to be at the TDC location with respect
to the initial surface mesh state. However, unlike a typical four-stroke setup where the piston “stretches”
the liner mesh from TDC to BDC, elongating the cylinder wall as it travels, a two-stroke piston must
“squeeze” an existing piston skirt mesh as it travels from TDC to BDC. Since Forte only translates vertices
located strictly on the piston, the skirt mesh must be capable of being compressed without causing
the surface topology to become invalid. This means the piston skirt should contain no intermediate
vertices in the space between BDC and TDC.
Tip
Thus, the easiest way to create piston geometry compatible with two-stroke motion is to
generate the surface mesh with the piston in the BDC position and translate the piston to
TDC after importing the geometry into Forte.
Figure 2.22: BDC position translated to TDC (p. 38), demonstrates the geometry in the initial configuration
on the left, and after translating the piston to TDC on the right. The skirt mesh lines are shown. Note
that no intermediate vertices exist between TDC and BDC positions so this skirt mesh can compress
without upsetting the topology of the surface mesh.
Best Practices
Figure 2.22: BDC position translated to TDC
When defining the piston boundary condition, the Motion Type parameter should be set to Sliding
Interface. In the subpanel associated with this selection, select the outer cylinder wall and inner wall
that define the piston skirt as the stationary and sliding surfaces in the selection control. The piston
skirt should be given its own boundary condition separate from the piston, and it should be defined
with Wall Motion unchecked.
Figure 2.23: Motion Type set as Sliding Interface
2.4.3. Surface Mesh Topology and Other Considerations

Pay special attention to the orientation of the surface normals of the piston, skirt, and walls. The direction
of normal vectors should be as shown in Figure 2.24: Directions of normal vectors (p. 39). This image
shows the most common topology configuration for two-stroke cases, but other topologies are possible.
Figure 2.24: Directions of normal vectors
For example, the piston can be defined as a completely enclosed surface anchored at the bottom of
the cylinder (with a gap between piston bottom and lower with similar constraints as the skirt-liner gap
described previously), with the top of the piston defined as the sole moving surface, as shown in Fig-
ure 2.25: Geometry with enclosed piston (p. 40)
Best Practices
Figure 2.25: Geometry with enclosed piston
It’s also possible to define an entirely moveable piston, in which case all surfaces are defined as part of
the piston boundary condition.
Figure 2.26: Geometry with moveable piston
2.4.4. Runtime Port Open/Close Considerations

Due to the interaction of the solver time step with the piston skirt and cylinder wall location, the precise
timing of when a port opens or closes may vary slightly from cycle to cycle or run to run. This can be
minimized by refining the geometry near the port openings. A feature angle refinement control set to
¼ mesh size is often a good choice.

ANSYS Forte Best Practices

Uploaded by

Copyright:

Available Formats

You might also like

ANSYS Forte Best Practices

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

ANSYS Forte Best Practices

Uploaded by

Copyright:

Available Formats

Forte Best Practices

ANSYS, Inc. Release 2019 R1

© 2018 ANSYS, Inc. Unauthorized use, distribution or duplication is prohibited.

U.S. Government Rights

Published in the U.S.A.

2.3.5.3. Output Controls ............................................................................................................ 32

• Diesel engine simulation

• Automatic mesh generation (AMG)

• Two-stroke engine simulation

2.1. Operating Environment

Recommended configurations for ANSYS Forte installations:

• Memory: 6 GB per core.

• A large hard disk is required for saving solution files.

2.2. DI Diesel Engine Case (Typically, Sector Mesh)

2.2.1.1. Sector Mesh Generation

2.2.1.1.1. Bowl Profile

2.2.1.1.2. Topology Selection, Mesh Control Points, and Size Parameters

2.2.1.2. Evaluating Initial (IVC) Gas Composition

2.2.1.2.1. Initial (IVC) NOx for High EGR Cases

2.2.1.3. Fuel Mass for Direct Injection Cases

Injected fuel mass [mg/cycle/cylinder] = fuel [mg/stroke/cylinder] * <number of strokes(4 or 2)>

2.2.2. Basic Calibration Procedures

2.2.2.2. Injection Timing

2.2.3. Advanced Calibration

Table 2.1: Parameters providing secondary impacts on engine behavior

Typical observed effect of

M KH Time 20–80 40 Delays spray -- -- Delayed;

2.2.4. Comparing to Experimental Data

1. Piston position, is defined as follows:

2. Volume can then be calculated as:

3. At the first CA point: dQ=0

4. At CA point n, where n = 1, .., last:

2.2.4.2. Importing Data Arrays into Forte Visualizer

2.2.4.3. Review Spray Behavior

2.2.4.4. Emissions and Performance Data

• IMEP (Indicated Mean Effective Pressure)

1. ISFC (Indicated Specific Fuel Consumption)

2.2.4.4.1. Conversion to Mass per Time Units

Emissions [g/h] = Emissions [g/kg-fuel]

* <number of cylinders> [cylinder/engine]

where Revolution_per_cycle is 2 for a four-stroke engine and 1 for a two-stroke engine.

2.3. Setting Up Automatic Mesh Generation Cases With Valve Motion

• The surface mesh file should have the piston at TDC.

Visualizer. Use the Launcher Preferences in the Forte Launcher.

2.3.1. Read In and Process the Surface Mesh

2. Configure the mesher to CFD, Fluent, and specify these options:

Physics Preference = CFD

Solver Preference = Fluent

Automatic Mesh Based Defeaturing = ON

Other settings may also be adjusted for good results.

4. Go to File > Export and save the .mesh file.

2.3.1.2. Splitting the Geometry

2. Exhaust port wall

3. Intake port wall

6. Piston surface (usually including bowl)

If you get a mesh-generation failure during preview, an incorrect surface normal is a

2.3.2. Set Up Mesh Size and Mesh Refinement

2.3.2.2. Global Mesh Size

2.3.2.3. Mesh Refinement