Working with a variety of users with different levels of CFD (Computational Fluid Dynamics) backgrounds, I have to admit that Fluent meshing used to be a challenging and confusing task for beginners and even intermediate users.
Ansys has addressed this challenge by redesigning the Fluent user interface to provide a task-based workflow for meshing that enables engineers to do more and solve more complex problems than ever before in less time. The new Fluent task-based workflow streamlines the user experience by providing a single window that offers only relevant choices and options and prompts the user with best practices that deliver better simulation results.
Best practices are embedded into the workflow in the form of defaults and messages to the user. This reduces the amount of training required to start using the software and makes it easier for occasional users to return to the software.
How to Mesh Watertight CFD Geometry in the New Ansys Fluent Task-based Workflow
In order to use this workflow, you need a relatively clean watertight solid and/or fluid regions that can be meshed by surface meshing and then volume filling (no wrapping required.) Geometry can consist of single or multiple bodies.
Going through the task-based workflow is straightforward. You are presented with several steps, like:
Describe geometry. (Fluid and/or solid)
Capping. (If you are creating an internal flow volume, then the capping tools in Fluent makes extraction easy)
Volume meshing. (If you wish to use the latest Mosaic meshing technology, select “Poly-hexcore”)
Now, click on “Switch-to-Solution,” to bring the mesh into a familiar Fluent interface.
Fault-Tolerant Workflow for Ansys Fluent Meshing Wraps and Seals Leaks and Gaps
Sometimes CFD simulations contain dirty, non-watertight geometries. For instance, 3D scanned or manufacturing geometry files. These geometries may contain missing faces, gaps, holes, overlaps, and other issues. As a result, they require extensive cleanup before simulation.
To overcome this obstacle, Ansys offers a new Fluent meshing workflow that wraps dirty geometry without cleanup.
The workflow for non-watertight geometry offers distinct advantages over other meshing technologies such as:
Users can perform CAD-level changes to any geometry or assembly, including dragging and dropping objects from the CAD model into the simulation model.
Leakages and overlaps:
The fault-tolerant workflow seals leakages caused by gaps and misalignments between solid bodies. This significantly reduces the manual efforts required to clean up geometry.
STL file input
The workflow can create fluid regions directly from STL files or scanned data. This eliminates the need to convert STL files into solid geometry for the biomedical, oil and gas, automotive and other industries.
There are a few important improvements both in Watertight meshing (WTM) and Fault-Tolerant meshing (FTM) workflows in the 2020R2 release.
FTM/WTM: Wild card selection in lists
The Meshing Workflows now have an option to use a persistent Wildcard string for selecting labels or zones. This is in addition to the Filter Text option previously available. The new Use WildCard option stores the wildcard string itself in recorded workflows instead of an explicit list of locations so that when they are played back with new geometries, the matching will be performed again and pick up any matching zones/labels that were not in the earlier geometry.
WTM: Support of Region-specific Sizing
You can specify region-specific Max Size and Growth Rates during the Volume Meshing task. If you enable Region-based Sizing, Fluent will compute default sizing specifications for each region. These can then be adjusted as required for each region.
WTM: Start From Imported Surface Mesh
This is useful if you have an established surface-meshing workflow or if you already have a mesh generated (perhaps from another preprocessor or an existing Fluent case) and want to use that as a starting point for Fluent meshing. Once you import the surface mesh you have the option of using it as it is, or selectively adding additional Local Size controls and/or remeshing particular surfaces as needed.
FTM: Continuous prism layers for Poly and Poly-Hexcore for Fluids
For the Fault-Tolerant Meshing Workflow you can now create continuous prism layers without stair-stepping within poly and poly-hexcore fluid regions. Note that this will apply in all zones of the region.
WTM: Support of Local Sizing on Labeled Edges
Once you have labeled the edges, you can select Edge Size in Add Local Sizing to prescribe a target size on the selected edge(s).
In addition to explicit modeling of finite arrays in Ansys HFSS, there are three other methods based on using unit cells. To learn more about the unit cell, please see part 1 of this blog.
In part 2, I will introduce and compare these 3 methods (1) Finite array defined using the array setting in unit cell, (2) Finite Array using Domain Decomposition Method (FADDM), (3) 3D component arrays (Figure 1). Please note that that the method 3 requires HFSS 2020R1 or newer.
Figure 1. (a) Unit cell, (b) FADDM, (c) 3D component array.
Finite Array using Unit Cell
After defining a unit cell (Figure 1a), you may simply define the number of elements, the spacing between them, and the scan angle. The assumption is that there is no mutual coupling, every element has the same radiation pattern and the same excitation. This is a good approximation for large arrays (10×10 or larger). This method may not be accurate enough in some cases, for example for a small number of elements; and when the antenna elements have main beams toward the angles that are close to the plane of the array and toward the other elements, causing a higher level of mutual coupling. However, smaller arrays won’t require as large of a compute resource as a large array.
The advantage of this method is its simulation speed. It requires the minimum memory and time to provide a quick array simulation. To define the array (after running the analysis for unit cell), right-click on Radiation from the Project Manager window. Select Antenna Array Setup, and then Regular Array Setup. In the Antenna Array Setup under Regular Array type, define the location of the first cell, the direction, the distance between the cells and number of cells in each direction.
Figure 2. Steps to define a finite array, (a) Antenna array setup, (b) & (c) Regular array setup.
Finite Array using Domain Decomposition Method
General Domain Decomposition (DDM) for a single domain provides a way to reduce memory requirement, however, it does not reduce the meshing time for large explicit arrays. Using Finite Array DDM (FADDM) addresses this shortcoming. The FADDM bypasses the adaptive meshing stage by duplicating the mesh that was generated for a unit cell. While the unit cell is used to create the mesh, the assumption of uniform excitation is no longer present. Each element in FADDM can have different magnitude and phase and is individually modeled, however, the mesh created in the unit cell is used to generate the overall mesh, therefore, no CPU time is spent on generating the mesh. This can be seen in Figure 3, by linking the mesh of the unit cell to the FADDM, the mesh is copied, and no mesh refinement will be needed. You may compare it with the explicit array of the same size (Figure 3(c)) where the entire array has to be meshed and mesh refinement will be necessary for adaptive meshing. This can be a huge simulation time saving when the array size is large.
Figure 3. (a) Mesh from a unit cell, (b) mesh linked to FADDM, (c) explicit array mesh.
To create FADDM from unit cell, create a new HFSS design. Then copy the unit cell into the new design. In the Project Manager window, right-click on Model and choose Create Array, as shown in Figure 4. This opens the window that allows user to define the number of elements of the array along the lattice directions. By selecting “Active Cells” tab, user can define where active, passive and padding cells are located. This gives the user a means of creating different lattice shapes.
Please note that padding cells defined in the General tab represent the size of vacuum buffer surrounding the array. They are not visible to the users but are included in the FADDM simulation. The same mesh from the unit cell simulation is duplicated to padding cells (Figure 5). It is also possible to add padding cells in the Active Cells tab, and those cells are also invisible, but can be used to create the array lattice of the desired shape (Figure 6).
Figure 4. The steps to create a DDM array, the Padding Cells are used to create the vacuum box and are invisible to the user.
Figure 5. The FADDM needs a padding cell to create a vacuum box around the design. The padding cells are invisible to the user.
Figure 6. (a) Padding cells can be used to create a lattice, (b) the lattice created does not show the padding cells.
The next step is to link the mesh to the unit cell. First, an analysis setup should be created. Choose Advanced Solution Setup. In the Driven Solution Setup General tab, reduce the number of Maximum Number of Passes to 1, as shown in Figure 7(a), then choose Advanced tab and click on Import Mesh (Figure 7(b)). Click on Setup Link. This is to link the simulation to the mesh of a unit cell. There are two steps needed here. First, choose the file or design that contains the mesh information (Figure 7(c)), second is to map the variables Figure 7(d). The last step to setting up the analysis is selecting Advanced Mesh Operation tab and selecting “Ignore mesh operation in target design” (Figure 7(d)). Now the array is ready and simulation can be run. You notice that adaptive meshing goes to only one pass. If in Setup Link window the option of “Simulate source design as needed” is checked (Figure 7(c)), then if a design variable that affects the geometry is changed, the meshing of the unit cell is repeated as needed. After the simulation is completed the elements magnitude and phases can be changed as a post processing step by “Edit Sources” (right-click on Excitations). The source names provided in the edit sources is slightly different than an explicit array (Figure 8)
Figure 7. Different windows related to setting up a linked mesh in FADDM.
Figure 8. Edit sources gives the ability to change the magnitude and phase of each element.
To compare the run-time and array patterns an example of circular polarized microstrip patch antenna of a 5 ´ 5 element array is shown in Figure 9 and Figure 10. The differences can be seen at angles away from the broadside angle. This shows how the edge effects are ignored in the unit cell approximation. Table 1 shows the comparison of memory and runtime for the three methods.
Table 1. Comparing run time and memory needed for a 5 x 5 array, explicit array vs FADDM.
Elapsed Time (min:sec)
Figure 9. Comparison of the far-field patterns for LHCP (co-polarization) created using FADDM and unit cell array.
Figure 10. Comparison of the far-field patterns for RHCP (cross-polarization) created using FADDM and unit cell array.
Finally, we compare the co-polarization pattern with an explicit 5 x 5 element array for scan angles of 0 and 30 degrees in Figure 11 and Figure 12, respectively.
Figure 11. Comparison of far-field LHCP created by explicit array vs. FADDM.
Figure 12. Comparison of scanned far-field LHCP, scan angle of 30 degrees.
3D Component Array
In 2020R1 the option of 3D component array was added. This option provides a means of combining different unit cells in one array. The unit cells are defined and imported as 3D components. To create a 3D component unit cell, define each type of the cells in a separate HFSS design, run the analysis, then select all objects in the model. In the Model ribbon, click on Create 3D Component, assign a name (no spaces are allowed in the name), add any information you like to add such as owner, email, company, etc., then click OK. Once all the 3D component cells are created, create a new HFSS design for the 3D component FADDM.
The next step is to create Relative CS for each of the 3D component elements in the HFSS design that will contain the array. For this step you need to plan the array lattice ahead of time, so the components are placed in the proper locations (Figure 13). Overlap is not allowed.
The unit cells should have the followings:
Identical dimensions of the bounding boxes
Identical Primary/Secondary (Master/Slave) boundary on the unit cells.
The generation of FADDM is similar to single unit cell array, except that when you select Model->Create Array, the window will be shown as 3D Component Array Properties (Figure 14). After choosing the number of elements and the number of padding cells, the unit cells window (Figure 15) will give you options of choosing one of the 3D component unit cells for each location of the elements in the lattice. The cells can be color coded. In the example shown in Figure 16 there are 3 components, the blank cell, the vertical cell and the horizontal cell. The sources under Edit Sources window are also arranged based on the name of the 3D component cells. At this point there is no option of linking the mesh. Therefore, the number of passes for adaptive mesh should be set to a number that is appropriate for getting a convergence.
Figure 13. 3D component unit cells are arranged to create a 3D component array.
Figure 14. The 3D component array can be created the same way as creating FADDM using a unit cell.
Figure 15. The unit cells are color coded for easier lattice creation.
Figure 16. The result of lattice created using 3D component array.
Unit cell and Finite Array Domain Decomposition are excellent options for simulating large finite arrays within a reasonable runtime and memory requirements. The 3D component finite array is a nice added feature in 2020R1 that now provides a way to combine unit cells with different geometries in one array.
If you would like more information related to this topic or have any questions, please reach out to us at firstname.lastname@example.org.
In this article, I would like to introduce some new features added to the Ansys Electronics Solution 2020R2 release called SPISim. Since this is a new tool, I’ll focus on describing its capabilities as well as some possible applications.
What is SPISim?
Signal, Power Integrity and Simulation (SPISim) focuses on system-level and on-chip SI/PI modeling, simulation, and analysis. The tool presents a variety of different features, which are split on separate modules shown below.
Let us look at each module individually and highlight the key functionality.
There are 2 main Modules VPro and MPro. All the other features (sub-modules) are split between these main two.
This is a versatile GUI for viewing waveforms. It supports a wide variety of formats including .tr0, .ac0, .ibis, .csv, .mat, .raw, .snp, .citi, and more. Besides simple viewing capabilities, VPro can also be used for waveform analysis:
Overshoot and Undershoot (for Peaks and Valleys)
Pulse to PDA
DPro Unit (VPro Module)
DPro (short for DDR Pro) provides comprehensive DDR related post-processing analysis. Key functionalities of this tool:
Batch mode of processing one or more waveform files
Support of multiple receiver processing
Built-in and customizable derating table and derating processing
Built-in 100+ measurement functions for typical DDR signal analysis
Results cross-probing and show problematic location automatically
The feature is organized in a wizard-like style. The user simply needs to fill out information in 6 tabs and click the ‘Run’ button. Overall, it is very intuitive to use, but like any new features, there is a learning curve for a new user.
TPro Unit (VPro Module)
It provides comprehensive transmission line related modeling, analysis, post-processing, and viewing capabilities. Here are several main functionalities offered by this add-on:
Comprehensive stackup planner to model t-lines’ performance in different stackup configurations
Advanced t-line modeling viewer for rapid analysis such as impedance, crosstalk, or propagation delay analysis
A table viewer for RLCG frequency content
What-if analysis for quick impedance/crosstalk calculation, and data processing, such as trimming and merging of frequency points
Batch mode processing and measurements for one or more t-line model files, result is a plain .csv file ready for further modeling or analysis
This feature helps the user to run pre-layout ‘what-if’ analysis. Both ‘transmission line analyzer’ and ‘layer stackup planner’ give the user a flexible way of understanding potential design constrains and guidelines.
SProUnit (VPro Module)
This module is similar to TPro in a sense of the capabilities. However, it is directed to view and analyze S-parameters instead of tabular transmission line data. Also, in contrast to TPro, this feature has a separate tab ‘S-Param’ with all the features listed there.
Here are major capabilities of SPro:
Advanced s-parameter viewer for speedy analysis such TDR/TDR or PDA analysis
Table viewer for frequency content; export s-parameter data to matlab .mat format and more
20+ advanced analysis functions such as mixed-mode conversion, cascading and renormalization
Batch mode processing and measurements for one or more s-parameter files
Support customizable s-parameter reporting generation for lab automation and beyond
Besides the conceptual similarities with the TPro, S-parameter’s waveform viewer based on VPro waveform viewer. Therefore, all operations available in VPro can also be found in S-parameter viewer.
This tool allows the user to generate a signal and use it in a future analysis. The generator offers wide variety of signal patterns (such as PRBS, Pulse, Sine, Square, Sawtooth etc) in combination with the PAM4 and NRZ modulation schemes. The user needs only to specify parameters for the signal and then create it.
This simple, but very powerful feature helps to save time for the engineer.
By definition, MPro is a modeling unit, which helps the user to work with the data. However, modeling can mean different things. The main advantage of MPro is providing the user with the simple environment for data manipulation. Here are all main functionalities of this core module:
Table data processing: combine, extract, summarize statistically, etc
Plan sampling with design of experiments, full factorial, Monte Carlo, etc
Simulate or collect data using customizable scripts, supporting multi-CPU/multi-thread
Visualize data in statistical, 2D or 3D plots
Model data using response surface modeling, neural network (feed forward and radial basis), etc
Optimization using linear, nonlinear, or genetic algorithm methods
BPro IBIS and AMI Unit (MPro Module)
BPro is one unit, however in this description I have purposefully separated it into two – BPro IBIS and BPro AMI, because the functionality of BPro is very broad. It is easier to focus on a one thing at a time.
Generally, BPro brings comprehensive IBIS related modeling, analysis, post-processing, and viewing capabilities to user. In more detail:
Has an inspector to view IBIS model’s textual content and visualize various waveform/current table easily. Tool also allows manual editing of model data with a simple mouse click and drag
Built-in advanced IBIS model generation flow from either scratch or existing simulation data. Tool will guide user from modeling setup, spice decks generations, simulation, modeling, syntax checking with golden parser, validation to final figure of merits (FOM) reporting
Support batch mode generations of performance reports for one or more model files. Results are in .csv file format and can be used for further analysis
IBIS model generation from Spec. or data sheet without performing any simulation. Generated model will also have two sets of waveforms under different loading conditions
Under ‘IBIS’ menu tab, the user will find separate sets of commands for both IBIS and AMI, as well as commands for IBIS-AMI in general.
This new addition to Ansys Electronics Solution brings a very wide variety of features to engineers. All Waveform Viewer, Signal Generator, IBIS-AMI modeling, DDR analysis, Data optimization, and Transmission line planner are united under one tool – SPISim. We can launch this tool either from within Ansys 3D Layout or SIwave, and, in 2020R2, is accessible through the Electronics Enterprise license.
Editor’s Note: The following PowerPoint is from one of PADT’s inhouse experts on linear dynamics, Alex Grishin.
One of the most valuable results that can come from a harmonic response analysis is the predicted alternating stresses in the part. This feeds fatigue and other downstream calculations as well as predicting maximum possible values. Becasue of the math involved, calculating derived stresses, like Principal Stresses can be done several ways. This post shows how Ansys Mechanical does it and offers an alternative that is considered more accurate for some cases.
Even before finishing my undergraduate degree in Mechanical
Engineering at the University of Colorado, Boulder in 2010, I had an interest in
furthering my education. The decision I had at that point was whether the next
step would be a graduate degree on the technical side or something more like an
MBA. I would end up with the chance to study at the University of Denver (DU),
focusing on Computational Fluid Dynamics (CFD), and if that field does not make
it clear, my first stint in grad school was technical.
At DU, we sourced our Ansys simulation software from a
company called, you guessed it, PADT. After finishing this degree, and while
working at PADT, the desire to further my education cropped up again after
seeing the need for a well-rounded understanding of the technical and
business/management side of engineering work. After some research, I decided
that a Master’s in Engineering Management program made more sense than an MBA,
and I started the program back at my original alma mater, CU Boulder.
Throughout the program, I would find myself using the skills
I was learning during lectures immediately in my work at PADT. It is difficult
to boil down everything learned in a 10-course program to one skill that is
used most often, and as I think about it, I think that what is used most
frequently is the new perspective, the new lens through which I can now view
situations. It’s taking a step back from the technical work and viewing a given
project or situation from a perspective shaped by the curriculum as a whole
with courses like EMEN 5020 – Finance and Accounting for Engineers, EMEN
5030/5032 – Fundamentals/Advanced Topics of Project Management, EMEN 5050 –
Leading Oneself, EMEN 5080 – Ethical Decision Making, EMEN 5500 – Lean and
Agile Management, and more. It is the creation of this new perspective that has
been most valuable and influential to my work as an engineer and comes from the
time spent completing the full program.
As I write this from home during the Covid-19 crisis, I want to assure you that PADT is conscious of many others working from home while using Ansys software as well. We’re trying to help with those who may be struggling with certain types of models. In this posting, I’ll talk about a contact convergence problem in Ansys Mechanical. I’ll discuss steps we can take to identify the problem and overcome it, as well as a simple setting to make which dramatically helped in this case.
The geometry in use here is a fairly simple assembly from an
old training class. It’s a wheel or
roller held by a housing, which is in turn bolted to something to hold it in
The materials used are linear properties for structural
steel. The loading consists of a bearing
load representing a downward force applied by some kind of strap or belt looped
over the wheel, along with displacement constraints on the back surfaces and
around the bolt holes, as shown in the image below. The flat faces on the back side have a
frictionless support applied (allows in plane sliding only), while the circular
faces where bolt heads and washers would be are fully constrained with fixed
As is always the case in Ansys Mechanical, contact pairs are created wherever touching surfaces in the assembly are detected. The default behavior for those contact pairs is bonded, meaning the touching surfaces can neither slide nor separate. We will make a change to the default for the wheel on its shaft, though, changing the contact behavior from bonded to frictional. The friction coefficient defined was 0.2. This represents some resistance to sliding. Unlike bonded contact in which the status of the contact pair cannot change during the analysis, frictional contact is truly nonlinear behavior, as the stiffness of the contact pair can change as deflection changes.
This shows the basic contact settings for the frictional
At this point, we attempt a solve. After a while, we get an error message stating, “An internal solution magnitude limit was exceeded,” as shown below. What this means is that our contact elements are not working as expected, and part of our structure is trying to fly off into space. Keep in mind in a static analysis there are no inertia effects, so an unconstrained body is truly unconstrained.
At this point, the user may be tempted to start turning
multiple knobs to take care of the situation.
Typical things to adjust for contact convergence problems are adding
more substeps, reducing contact stiffness, and possibly switching to the
unsymmetric solver option when frictional contact is involved. In this case, a simple adjustment is all it
takes to get the solution to easily converge.
Another thing we might do to help us is to insert a Contact
Tool in the Connections branch and interrogate the initial contact status:
This shows us that our frictional contact region is actually
not in initial contact but has a gap.
There are multiple techniques available for handling this situation,
such as adding weak springs, running a transient solution (computationally
expensive), starting with a displacement as a load and then switching to a
force load, etc. However, if we are
confident that these parts actually SHOULD be initially touching but are not
due to some slop in the CAD geometry, there is a very easy adjustment to handle
The Simple Adjustment That Gets This Model to Solve
Knowing that the parts should be initially in contact, one
simple adjustment is all that is needed to close the initial gap and allow the
simulation to successfully solve. The
adjustment is to set the Interface Treatment in the Contact Details for the
contact region in question to Adjust to Touch:
This change automatically closes the initial gap and, in
this case, allows the solution to successfully solve very quickly.
For your models, if you are confident that parts should be
in initial contact, you may also find that this adjustment is a great aid in
closing gaps due to small problems in the CAD geometry. We encourage you to test it out.
Ansys optiSLang is one of the newer pieces of software in the Ansys toolkit that was recently acquired along with the company Dynardo. Functionally, optiSLang provides a flexible top-level platform for all kinds of optimization. It is solver agnostic, in that as long as you can run a solver through batch files and produce text readable result files, you can use said solver with optiSLang. There are also some very convenient integrations with many of the Ansys toolkit solvers in addition to other popular programs. This includes AEDT, Workbench, LS-DYNA, Python, MATLAB, and Excel, among many others.
While the ultimate objective is often to simply minimize or maximize a system output according to a set of inputs, the complexity of the problem can increase dramatically by introducing constraints and multiple optimization goals. And of course, the more complicated the relationships between variables are, the harder it gets to adequately describe them for optimization purposes.
Much of what optiSLang can do is a result of fitting the input data to a Metamodel of Optimal Prognosis (MOP) which is a categorical description for the specific metamodels that optiSLang uses. A user can choose one of the included models (Polynomial, Moving Least Squares, and Ordinary Kriging), define their own model, and/or allow optiSLang to compare the resulting Coefficients of Prognosis (COP) from each model to choose the most appropriate approach.
The COP is calculated in a similar manner as the more common COD or R2 values, except that it is calculated through a cross-validation process where the data is partitioned into subsets that are each used only for the MOP calculation or the COP calculation, not both. For this reason, it is preferred as a measure for how effective the model is at predicting unknown data points, which is particularly valuable in this kind of MOP application.
This whole process really shows where optiSLang’s functionality shines: workflow automation. Not only does optiSLang intelligently select the metamodel based on its applicability to the data, but it can also apply an adaptive method for the improvement of the MOP. It will suggest an automatic sampling method based on the number of problem variables involved, which can then be applied towards refining either the global COP or the minimum local COP. The automation of this process means that once the user has linked optiSLang to a solver with appropriate inputs/outputs and defined the necessary run methodology for optimization, all that is left is to click a button and wait.
As an example of this, we will run through a test case that utilizes the ability to interface with Ansys EDT Icepak.
Figure 1: The EDT Icepak project model.
For our setup, we have a simple board mounted with bodies representative of 3 x 2 watt RAM modules, and 2 x 10 watt CPUs with attached heatsinks. The entire board is contained within an air enclosure, where boundary conditions are defined as walls with two parametrically positioned circular inlets/outlets. The inlet is a fixed mass flow rate surface and the outlet is a zero-pressure boundary. In our design, we permit the y and z coordinates for the inlet and outlet to vary, and we will be searching for the configuration that minimizes the resulting CPU and RAM temperatures.
The optiSLang process generally follows a series of drag-and-drop wizards. We start with the Solver Wizard which guides us through the options for which kind of solver is being used: text-based, direct integrations, or interfaces. In this case, the Icepak project is part of the AEDT interface, so optiSLang will identify any of the parameters defined within EDT as well as the resulting report definitions. The Parametric Solver System created through the solver wizard then provides the interfacing required to adjust inputs while reading outputs as designs are tested and an MOP is generated.
Figure 2: Resulting block from the Solver wizard with parameters read in from EDT.
Once the parametric solver is defined, we drag and drop in a
sensitivity wizard, which starts the AMOP study. We will start with a total of 100 samples; 40
will be initial designs, and 60 will be across 3 stages of COP refinement with
all parameter sets sampled according to the Advanced Latin Hypercube Sampling
Figure 3: Resulting block from the Sensitivity wizard with Advanced Latin Hypercube Sampling.
The results of individual runs are tabulated and viewable as the study is conducted, and at the conclusion, a description of the AMOP is provided with response surfaces, residual plots, and variable sensitivities. For instance, we can see that by using these first 100 samples, a decent metamodel with a COP of 90% is generated for the CPU temperature near the inlet. We also note that optiSLang has determined that none of the responses are sensitive to the ‘y’ position of the outlet, so this variable is automatically freed from further analysis.
Figure 4: MOP surface for the temperature of Chip1, resulting from the first round of sampling.
If we decide that this CoP, or that from any of our other outputs, is not good enough for our purposes, optiSLang makes it very easy to add on to our study. All that is required is dragging and dropping a new sensitivity wizard onto our previous study, which will automatically load the previous results in as starting values. This makes a copy of and visually connects an output from the previous solver block to a new sensitivity analysis on the diagram, which we can then be adjusted independently.
For simplicity and demonstration’s sake, we will add on two more global refinement iterations of 50 samples each. By doing this and then excluding 8 of our 200 total samples that appear as outliers, our “Chip1” CoP can be improved to 97%.
Figure 5: A refined MOP generated by including a new Sensitivity wizard.
Now that we have an MOP of suitable predictive power for our outputs of interest, we can perform some fast optimization. By initially building an MOP based on the overall system behavior, we are now afforded some flexibility in our optimization criteria. As in the previous steps, all that is needed at this point is to drag and drop an optimization wizard onto our “AMOP Addition” system, and optiSLang will guide us through the options with recommendations based on the number of criteria and initial conditions.
In this case, we will define three optimization criteria for thoroughness: a sum of both chip temperatures, a sum of all RAM temperatures, and an average temperature rise from ambient for all components with double weighting applied to the chips. Following the default optimization settings, we end up with an evolutionary algorithm that iterates through 9300 samples in about 14 minutes – far and away faster than directly optimizing the Icepak project. What’s more, if we decide to adjust the optimization criteria, we’ll only need to rerun this ~14 minute evolutionary algorithm.
What we are most interested in for this example are the
resulting Pareto fronts which give us a clear view of the tradeoffs between
each of our optimization criteria. Each of the designs on this front can easily
be selected through the interface, and their corresponding input parameters can
Figure 6: Pareto front of the “Chipsum” and “TotalAve” optimization criteria.
Scanning through some of these designs also provides a very convenient way to identify which of our parameters are limiting the design criteria. Two distinct regions can be identified here: the left region is limited by how close we are allowing the inlet fan to be to the board, and the right region is limited by how close to the +xz corner of our domain the outlet vent can be placed. In a situation where these parameters were not physically constrained by geometry, this would be a good opportunity to consider relaxing parameter constraints to further improve our optimization criteria.
As it is, we can now choose a design based on this Pareto front to verify with the full solver. After choosing a point in the middle of the “Limited by outlet ‘z’” zone, we find that our actual “ChipSum” is 73.33 vs. the predicted 72.78 and the actual “TotalAve” is 17.82 vs. the predicted 17.42. For this demonstration, we consider this small error as satisfactory, and a snapshot of the corresponding Icepak solution is shown below.
Figure 7: The Icepak solution of the final design. The inlet vent is aligned with the outlet side’s heatsink, and the outlet vent is in the corner nearest the heatsink. Primary flow through the far heatsink is maximized, while a strong recirculating flow is produced around the front heatsink.
The accuracy of these results is of course dependent not only on how thoroughly we constructed the MOP, but also the accuracy of the 3D solution; creating mesh definitions that remain consistently accurate through parameterized geometry changes can be particularly tricky. Though, with all of this considered, optiSLang provides a great environment for not only managing optimization studies, but displaying the results in such a way that you can gain an improved understanding of the interaction between input/output variables and their optimization criteria.
Ansys Polyflow is a Finite Element CFD solver with unique capabilities that enable simulation of complex non-Newtonian flows seen in the polymer processing industry. In recent releases, Polyflow has included templates to streamline two of its most common use cases: blow molding and extrusion. Similarly, Ansys Discovery AIM offers a modern user interface that guides users through blow molding and extrusion workflows while still using the proven Polyflow solver under the hood. It is not uncommon for engineers to be unsure about which tool to pursue for their specific application. In this article, I will focus on the blow molding workflow. More specifically, I will point out three features in Polyflow that have not yet been incorporated into Discovery AIM:
The PolyMat curve fitting tool to derive
viscoelasticity model input parameters from test data
Automatic parison thickness mapping onto an
Ansys Mechanical shell mesh
Parison Programming to optimize parison
thickness subject to final part thickness constraints
Keep in mind that either tool will get the job done in most
applications, so let us first quickly review some of the core features of blow
molding simulations that are common to Polyflow and AIM:
Parison/Mold contact detection
3-D Shell Lagrangian automatic remeshing
Generalized Newtonian viscosity models
Temperature dependent and multi-mode integral viscoelastic
Time dependent mold pressure boundary conditions
or non-isothermal conditions
For demonstration purposes, I modeled a sweet submarine toy
in SpaceClaim. Unfortunately, I think it will float, but let’s move past that
Figure 1: Final Submarine shape (Left), Top View of Mold+ Parison (Top Left), Side View of Mold+Parison (Bottom Right)
At this point, you could
proceed with Discovery AIM or with Polyflow without any re-work. I’lll proceed
with the Polyflow Blow Molding workflow to point out the features currently
only available in Polyflow.
PolyMat Curve Fitting Tool
With the blow molding template, you can select whether to treat the parison as isothermal or non-isothermal and whether to model it as general Newtonian or viscoelastic. Suppose we would like to model viscoelasticity with the KBKZ integral viscoelastic model because we were interested in capturing strain hardening as the parison is stretched. The inputs to the KBKZ model are viscosity and relaxation times for each mode. If they are known, the user can input the values directly. This is possible in Discovery AIM as well. However, the PolyMat tool is unique to Polyflow. PolyMat is a built-in curve fitting tool that helps generate input parameters for the various viscosity model available in Polyflow using material data. This is particularly useful when you do not explicitly have the inputs for a viscoelastic model, but perhaps you have other test data such as oscillatory and capillary rheometry data. In this case I have with the loss modulus, storage modulus and shear viscosity for a generic high density polyethylene (HDPE) material. For this material, four modes are enough to anchor the KBKZ model to the data as shown below. We can then load the viscosity/relaxation time into Polyflow and continue.
Figure 2: Curve Fitting of G’(Ω),G’’(Ω),η() [Left], KBKZ Viscoelastic Model inputs (Right)
The main output of the simulation is the final parison
thickness distribution. For this sweet submarine, the initial parison thickness
is set to 3mm and the final thickness distribution is shown in the contour plot
Figure 3a: Animation of blow molding process
Figure 3b: Final Part Thickness Distribution
Thickness Mapping to Ansys Mechanical
The second Polyflow capability I’d like to point out is the
ability to easily map the thickness distribution onto an Ansys mechanical shell
mesh. You can map the thickness onto an Ansys Mechanical shell mesh by
connecting the polyflow solution component to a structural model in workbench
as shown below. The analogous work flow in AIM, would be to create a second
simulation for the structural analysis, but you would be confined to specifying
a constant thickness.
In Ansys Mechanical, the mapping comes through within the
geometry tree as shown below. The imported Data Transfer Summary is a good way
to ensure the mapping behaves as expected. In this case we can see that 100% of
the nodes were mapped and the thickness contours qualitatively match the
Polyflow results in CFD -Post.
Figure 5: Imported Thickness in Ansys Mechanical
Figure 6: Thickness Data Transfer Summary
A force is applied normal to front face of the sail and simulated in Mechanical. The peak stress and deformation are shown below. The predicted stresses are likely acceptable for a toy, especially since my toy is a sweet submarine. Nonetheless, suppose that I was interested in reducing the deformation in the sail under this load condition by thickening the extruded parison. A logical approach would be to increase the initial parison thickness from 3mm to 4mm for example. Polyflow’s parison programming feature takes the guesswork out of the process.
Figure 7: Clockwise from Top Left: Applied Load on Sail, Stress Distribution, total Deformation, Thickness Distribution
Parison programming is an iterative optimization work flow
within Polyflow for determining the extruded thickness distribution required to
meet the final part thickness constraints. To activate it, you create a new post
processor sub-task of type parison programming.
Figure 8: Parison Programming Setup
The inputs to the optimization are straight forward. The only inputs that you typically would need to modify are the direction of optimization, width of stripes, and list of (X,h) pairs. The direction of optimization is the direction of extrusion which is X in this case. If the extruder can vary parison thickness along “stripes” of the parison, then Polyflow can optimize each stripe thickness. The list of (X,h) pairs serves as a list of constraints for the final part thickness where X is the location on the parison along the direction of extrusion and h is the final part thickness constraint.
Figure 9: Thickness Constraints for Parison Programming
In our scenario, the X,h pairs form a piecewise linear thickness distribution to constrain the area around the sail to have a 3.5mm thickness and 2mm everywhere else. After the simulation, Polyflow will write a csv file with to the output directory containing the initial thickness for each node for the next iteration. You will need to copy over the csv file from the output directory of iteration N to the input directory of iteration N+1. The good news is the optimization converges within 3-5 iterations.
Figure 10: Defining the Initial Thickness for the Next Parison Programming Iteration
Polyflow will print the parison strip thickness distribution
for the next iteration in the .lst file. The plot below shows the thickness
distribution from the first 3 iterations. Note from the charts below that the
distribution converged by iteration 2; thus iteration 3 was not actually
simulated. The optimized parison thickness distribution is also plotted in the
contour plot below.
Figure 11: Optimized Parison Thickness (Top), Final Part Thickness (Bottom)
% of Elements At or Above Thickness Criteria
As a final check, we can evaluate how the modification to the parison thickness reduced the deformation of the submarine. The total deformation contour plot below confirms that the peak deformation decreased from 2mm to 0.8mm.
Figure 13: Total Deformation in Ansys Mechanical After Parison Programming
Ansys Discovery AIM is a versatile platform with an intuitive and modern user interface. While Aim has incorporated most of the blow molding simulation capabilities from Polyflow, some advanced functionality has not yet been brought into AIM. This article simulated the blow molding process of a toy submarine to demonstrate three capabilities currently only available in Polyflow: the PolyMat curve fitting tool, automatic parison thickness mapping to Ansys Mechanical, and parison programming. Engineers should consider whether any of these capabilities are needed in their application next time they are faced with the decision to create a blow mold simulation using Ansys Discovery AIM or Polyflow.
There are three main goals of the licensing changes in the latest release of ANSYS:
Deliver Ansys licensing using the FlexLM industry standard
Eliminate the Ansys licensing interconnect
Provide modular licensing options that are easier to understand
Finally – and this is the whopper (or Double Double if you’re an In-N-Out kind of analogy person) – this new licensing model eliminates the need for upgrading the Ansys License Manager with every software update. (please pause for shock recovery)
Why is this significant? Well, this was always a
sticking point for our customers when upgrading from one version to the next.
Here’s how that usually plays out:
Engineers eager to try out new features or overcome software defects, download the software and install it on their workstations.
Surprise – software throws an obscure licensing error.
Engineer notifies IT or Ansys Channel partner of issue.
After a few calls, maybe a screenshare or two, its determined that the license server needs to be upgraded.
The best-case scenario – IT or PADT Support can get it installed in a few minutes and engineer can be on his way.
The usual scenario – it will take a week to schedule downtime on the server and notify all stakeholders and the engineer is left to simmer on medium-low until those important safeguards are handled.
What does this all mean?
Starting in January 2020, all new Ansys keys
issued will be in the new format and will require upgrading to the 2020R1
License manager. This should be the last mandatory license server upgrade
for a while.
Your Ansys Channel Partner will contact you ahead
of your next renewal to discuss new license increments and if there are any
Your IT and Ansys support team will be
celebrating in the back office the last mandatory Ansys License Manager upgrade
for a while.
How to upgrade the Ansys License Manager?
Download the latest license manager through the
Ansys customer portal:
In this article I will cover a Voltage Drop (DC IR) simulation in SIwave, applying realistic power delivery setup on a simple 4-layer PCB design. The main goal for this project is to understand what data we receive by running DC IR simulation, how to verify it, and what is the best way of using it.
And before I open my tools and start diving deep into this topic, I would like to thank Zachary Donathan for asking the right questions and having deep meaningful technical discussions with me on some related subjects. He may not have known, but he was helping me to shape up this article in my head!
There are many different power nets present on the board
under test, however I will be focusing on two widely spread nets +1.2V and
+3.3V. Both nets are being supplied through Voltage Regulator Module (VRM),
which will be assigned as a Voltage Source in our analysis. After careful
assessment of the board design, I identified the most critical components for
the power delivery to include in the analysis as Current Sources (also known as
‘sinks’). Two DRAM small outline integrated circuit (SOIC) components D1 and D2
are supplied with +1.2V. While power net +3.3V provides voltage to two quad
flat package (QFP) microcontrollers U20 and U21, mini PCIE connector, and hex
Schmitt-Trigger inverter U1.
Figure 1 shows the ‘floor plan’ of the DC IR analysis
setup with 1.2V voltage path highlighted in yellow and 3.3V path highlighted in
Before we assign any Voltage and Current sources, we need to
define pin groups for all nets +1.2V, +3.3V and GND for all PDN component
mentioned above. Having pin groups will significantly simplify the reviewing
process of the results. Also, it is generally a good practice to start the DC
IR analysis from the ‘big picture’ to understand if certain component gets
enough power from the VRM. If a given IC reports an acceptable level of voltage
being delivered with a good margin, then we don’t need to dig deeper; we can
instead focus on those which may not have good enough margins.
Once we have created all necessary pin groups, we can assign
voltage and current sources. There are several ways of doing that (using wizard
or manual), for this project we will use ‘Generate Circuit Element on
Components’ feature to manually define all sources. Knowing all the components
and having pin groups already created makes the assignment very
straight-forward. All current sources draw different amount of current, as indicated
in our setting, however all current sources have the same Parasitic Resistance
(very large value) and all voltage source also have the same Parasitic
Resistance (very small value). This is shown on Figure 2 and Figure 3.
Note: The type of the current source ‘Constant Voltage’ or
‘Distributed Current’ matters only if you are assigning a current source to a
component with multiple pins on the same net, and since in this project we are
working with pins groups, this setting doesn’t make difference in final results.
For each power net we have created a voltage source on VRM and multiple current sources on ICs and the connector. All sources have a negative node on a GND net, so we have a good common return path. And in addition, we have assigned a negative node of both voltage sources (one for +1.2V and one for +3.3V) as our reference points for our analysis. So reported voltage values will be referenced to that that node as absolute 0V.
At this point, the DC IR setup is complete and ready for
Results overview and validation
When the DC IR simulation is finished, there is large amount
of data being generated, therefore there are different ways of viewing results,
all options are presented on Figure 4. In this article I will be
primarily focusing on ‘Power Tree’ and ‘Element Data’. As an additional source
if validation we may review the currents and voltages overlaying the design to
help us to visualize the current flow and power distribution. Most of the time
this helps to understand if our assumption of pin grouping is accurate.
First let’s look at the Power Tree, presented on Figure 5. Two different power nets were simulated, +1.2V and +3.3V, each of which has specified Current Sources where the power gets delivered. Therefore, when we analyze DC IR results in the Power tree format, we see two ‘trees’, one for each power net. Since we don’t have any pins, which would get both 1.2V and 3.3V at the same time (not very physical example), we don’t have ‘common branches’ on these two ‘trees’.
Now, let’s dissect all the information present in this power tree (taking in consideration only one ‘branch’ for simplicity, although the logic is applicable for all ‘branches’):
We were treating both power nets +1.2V and +3.3V as separate voltage loops, so we have assigned negative nodes of each Voltage Source as a reference point. Therefore, we see the ‘GND’ symbol ((1) and (2)) for each voltage source. Now all voltage calculations will be referenced to that node as 0V for its specific tree.
Then we see the path from Voltage Source to Current Source, the value ΔV shows the Voltage Drop in that path (3). Ultimately, this is the main value power engineers usually are interested in during this type of analysis. If we subtract ΔV from Vout we will get the ‘Actual Voltage’ delivered to the specific current source positive pin (1.2V – 0.22246V = 0.977V). That value reported in the box for the Current Source (4). Technically, the same voltage drop value is reported in the column ‘IR Drop’, but in this column we get more details – we see what the percentage of the Vout is being dropped. Engineers usually specify the margin value of the acceptable voltage drop as a percentage of Vout, and in our experiment we have specified 15%, as reported in column ‘Specification’. And we see that 18.5% is greater than 15%, therefore we get ‘Fail_I_&_V’ results (6) for that Current Source.
Regarding the current – we have manually specified the current value for each Current Source. Current values in Figure 2 are the same as in Figure 5. Also, we can specify the margin for the current to report pass or fail. In our example we assigned 108A as a current at the Current Source (5), while 100A is our current limit (4). Therefore, we also got failed results for the current as well.
As mentioned earlier, we assigned current values for each Current Source, but we didn’t set any current values for the Voltage Source. This is because the tool calculates how much current needs to be assigned for the Voltage Source, based on the value at the Current Sources. In our case we have 3 Current Sources 108A, 63A, 63A (5). The sum of these three values is 234A, which is reported as a current at the Voltage Source (7). Later we will see that this value is being used to calculate output power at the Voltage Source.
This option shows us results in the tabular representation. It lists many important calculated data points for specific objects, such as bondwire, current sources, all vias associated with the power distribution network, voltage probes, voltage sources.
Let’s continue reviewing the same power net +1.2V and the power distribution to CPU1 component as we have done for Power Tree (Figure 5). The same way we will be going over the details in point-by-point approach:
First and foremost, when we look at the information for Current Sources, we see a ‘Voltage’ value, which may be confusing. The value reported in this table is 0.7247V (8), which is different from the reported value of 0.977V in Power Tree on Figure 5 (4). The reason for the difference is that reported voltage value were calculated at different locations. As mentioned earlier, the reported voltage in the Power Tree is the voltage at the positive pin of the Current Source. The voltage reported in Element Data is the voltage at the negative pin of the Current Source, which doesn’t include the voltage drop across the ground plane of the return path.
To verify the reported voltage values, we can place Voltage Probes (under circuit elements). Once we do that, we will need to rerun the simulation in order to get the results for the probes:
Two terminals of the ‘VPROBE_1’ attached at the positive pin of Voltage Source and at the positive pin of the Current Source. This probe should show us the voltage difference between VRM and IC, which also the same as reported Voltage Drop ΔV. And as we can see ‘VPROBE_1’ = 222.4637mV (13), when ΔV = 222.464mV (3). Correlated perfectly!
Two terminals of the ‘VPROBE_GND’ attached to the negative pin of the Current Source and negative pin of the Voltage Source. The voltage shown by this probe is the voltage drop across the ground plane.
If we have 1.2V at the positive pin of VRM, then voltage drops 222.464mV across the power plane, so the positive pin of IC gets supplied with 0.977V. Then the voltage at the Current Source 0.724827V (8) being drawn, leaving us with (1.2V – 0.222464V – 0.724827V) = 0.252709V at the negative pin of the Current Source. On the return path the voltage drops again across the ground plane 252.4749mV (14) delivering back at the negative pin of VRM (0.252709V – 0.252475V) = 234uV. This is the internal voltage drop in the Voltage Source, as calculated as output current at VRM 234A (7) multiplied by Parasitic Resistance 1E-6Ohm (Figure 3) at VRM. This is Series R Voltage (11)
Parallel R Current of the Current source is calculated as Voltage 724.82mV (8) divided by Parasitic Resistance of the Current Source (Figure 3) 5E+7 Ohm = 1.44965E-8 (9)
Current of the Voltage Source report in the Element Data 234A (10) is the same value as reported in the Power Tree (sum of all currents of Current Sources for the +1.2V power net) = 234A (7). Knowing this value of the current we can multiple it by Parasitic Resistance of the Voltage Source (Figure 3) 1E-6 Ohm = (234A * 1E-6Ohm) = 234E-6V, which is equal to reported Series R Voltage (11). And considering that the 234A is the output current of the Voltage Source, we can multiple it by output voltage Vout = 1.2V to get a Power Output = (234A * 1.2V) = 280.85W (12)
In addition to all provided above calculations and explanations, the video below in Figure 7 highlights all the key points of this article.
By carefully reviewing the Power Tree and Element Data reporting options, we can determine many important decisions about the power delivery network quality, such as how much voltage gets delivered to the Current Source; how much voltage drop is on the power net and on the ground net, etc. More valuable information can be extracted from other DC IR results options, such as ‘Loop Resistance’, ‘Path Resistance’, ‘RL table’, ‘Spice Netlist’, full ‘Report’. However, all these features deserve a separate topic.
As always, if you would like to receive more information related to this topic or have any questions please reach out to us at email@example.com.
HFSS offers different method of
creating and simulating a large array. The explicit method, shown in Figure
1(a) might be the first method that comes to our mind. This is where you create
the exact CAD of the array and solve it. While this is the most accurate method
of simulating an array, it is computationally extensive. This method may be
non-feasible for the initial design of a large array. The use of unit cell (Figure
1(b)) and array theory helps us to start with an estimate of the array
performance by a few assumptions. Finite Array Domain Decomposition (or FADDM)
takes advantage of unit cell simplicity and creates a full model using the
meshing information generated in a unit cell. In this blog we will review the
creation of unit cell. In the next blog we will explain how a unit cell can be
used to simulate a large array and FADDM.
In a unit cell, the following assumptions are made:
The pattern of each element is identical.
The array is uniformly excited in amplitude, but not necessarily in phase.
Edge affects and mutual coupling are ignored
A unit cell works based on
Master/Slave (or Primary/Secondary) boundary around the cell. Master/Slave
boundaries are always paired. In a rectangular cell you may use the new Lattice
Pair boundary that is introduced in Ansys HFSS 2020R1. These boundaries are
means of simulating an infinite array and estimating the performance of a
relatively large arrays. The use of unit cell reduces the required RAM and
Primary/Secondary (Master/Slave) (or P/S) boundaries can be
combined with Floquet port, radiation or PML boundary to be used in an infinite
array or large array setting, as shown in Figure 3.
To create a unit cell with P/S boundary, first start with a single element with the exact dimensions of the cell. The next step is creating a vacuum or airbox around the cell. For this step, set the padding in the location of P/S boundary to zero. For example, Figure 4 shows a microstrip patch antenna that we intend to create a 2D array based on this model. The array is placed on the XY plane. An air box is created around the unit cell with zero padding in X and Y directions.
You notice that in this example
the vacuum box is larger than usual size of quarter wavelength that is usually used
in creating a vacuum region around the antenna. We will get to calculation of this
size in a bit, for now let’s just assign a value or parameter to it, as it will
be determined later. The next step is to define P/S to generate the lattice. In
AEDT 2020R1 this boundary is under “Coupled” boundary. There are two methods to
create P/S: (1) Lattice Pair, (2) Primary/Secondary boundary.
The Lattice Pair works best for
square lattices. It automatically assigns the primary and secondary boundaries.
To assign a lattice pair boundary select the two sides that are supposed to
create infinite periodic cells, right-click->Assign
Boundary->Coupled->Lattice Pair, choose a name and enter the scan angles.
Note that scan angles can be assigned as parameters. This feature that is introduced
in 2020R1 does not require the user to define the UV directions, they
are automatically assigned.
Primary/Secondary boundary is the same as what used to be called Master/Slave boundary. In this case, each Secondary (Slave) boundary should be assigned following a Primary (Master) boundary UV directions. First choose the side of the cell that Primary boundary. Right-click->Assign Boundary->Coupled->Primary. In Primary Boundary window define U vector. Next select the secondary wall, right-click->Assign Boundary->Couple->Secondary, choose the Primary Boundary and define U vector exactly in the same direction as the Primary, add the scan angles (the same as Primary scan angles)
Floquet Port and Modes Calculator
Floquet port excites and
terminates waves propagating down the unit cell. They are similar to waveguide
modes. Floquet port is always linked to P/S boundaries. Set of TE and TM modes
travel inside the cell. However, keep in mind that the number of modes that are
absorbed by the Floquet port are determined by the user. All the other modes
are short-circuited back into the model. To assign a Floquet port two major
steps should be taken:
Defining Floquet Port
Select the face of the cell that you like to assign the Floquet port. This is determined by the location of P/S boundary. The lattice vectors A and B directions are defined by the direction of lattice (Figure 7).
The number of modes to be
included are defined with the help of Modes Calculator. In the Mode Setup tab
of the Floquet Port window, choose a high number of modes (e.g. 20) and click
on Modes Calculator. The Mode Table Calculator will request your input of
Frequency and Scan Angles. After selecting those, a table of modes and their
attenuation using dB/length units are created. This
is your guide in selecting the height of the unit cell and vaccume box. The
attenation multiplied by the height of the unit cell (in the project units,
defined in Modeler->Units) should be large enough to make sure the modes are
attenuated enough so removing them from the calcuatlion does not cause errors.
If the unit cell is too short, then you will see many modes are not attenuated
enough. The product of the attenuatin and height of the airbox should be at
least 50 dB. After the correct size for the airbox is calcualted and entered,
the model with high attenuation can be removed from the Floquet port
The 3D Refinement tab is used to control the inclusion of the modes in the 3D refinement of the mesh. It is recommended not to select them for the antenna arrays.
In our example, Figure 8 shows
that the 5th mode has an attenuation of 2.59dB/length. The height of
the airbox is around 19.5mm, providing 19.5mm*2.59dB/mm=50.505dB attenuation
for the 5th mode. Therefore, only the first 4 modes are kept for the
calculations. If the height of the airbox was less than 19.5mm, we would need to
increase the height so accordingly for an attenuation of at least 50dB.
A simpler alternative for Floquet port is radiation boundary. It is important to note that the size of the airbox should still be kept around the same size that was calculated for the Floquet port, therefore, higher order modes sufficiently attenuated. In this case the traditional quarter wavelength padding might not be adequate.
Perfectly Matched Layer
Although using radiation boundary
is much simpler than Floquet port, it is not accurate for large scan angles. It
can be a good alternative to Floquet port only if the beam scanning is limited
to small angles. Another alternative to Floquet port is to cover the cell by a
layer of PML. This is a good compromise and provides very similar results to
Floquet port models. However, the P/S boundary need to surround the PML layer
as well, which means a few additional steps are required. Here is how you can
Reduce the size of the airbox* slightly, so after adding the PML layer, the unit cell height is the same as the one that was generated using the Modes Calculation. (For example, in our model airbox height was 19mm+substrte thickness, the PML height was 3mm, so we reduced the airbox height to 16mm).
Choose the top face and add PML boundary.
Select each side of the airbox and create an object from that face (Figure 10).
Select each side of the PML and create objects from those faces (Figure 10).
Select the two faces that are on the same plane from the faces created from airbox and PML and unite them to create a side wall (Figure 10).
Then assign P/S boundary to each pair of walls (Figure 10).
*Please note for this method, an auto-size “region” cannot be used, instead draw a box for air/vacuum box. The region does not let you create the faces you need to combine with PML faces.
The advantage of PML termination over Floquet port is that it is simpler and sometimes faster calculation. The advantage over Radiation Boundary termination is that it provides accurate results for large scan angles. For better accuracy the mesh for the PML region can be defined as length based.
Seed the Mesh
To improve the accuracy of the PML model further, an option is to use length-based mesh. To do this select the PML box, from the project tree in Project Manager window right-click on Mesh->Assign Mesh Operation->On Selection->Length Based. Select a length smaller than lambda/10.
Scanning the Angle
In phased array simulation, we are mostly interested in the performance of the unit cell and array at different scan angles. To add the scanning option, the phase of P/S boundary should be defined by project or design parameters. The parameters can be used to run a parametric sweep, like the one shown in Figure 12. In this example the theta angle is scanned from 0 to 60 degrees.
Comparing PML and Floquet Port with Radiation Boundary
To see the accuracy of the
radiation boundary vs. PML and Floquet Port, I ran the simulations for scan
angles up to 60 degrees for a single element patch antenna. Figure 13 shows
that the accuracy of the Radiation boundary drops after around 15 degrees
scanning. However, PML and Floquet port show similar performance.
To compare the accuracy, we can also check the S parameters. Figure 14 shows the comparison of active S at port 1 for PML and Floquet port models. Active S parameters were used since the unit cell antenna has two ports. Figure 15 shows how S parameters compare for the model with the radiation boundary and the one with the Floquet port.
The unit cell definition and options on terminating the cell
were discussed here. Stay tuned. In the next blog we discuss how the unit cell
is utilized in modeling antenna arrays.
As systems become more complex, the introduction and
adoption of detailed Multiphysics / Multidomain tools is becoming more
commonplace. Oftentimes, these tools serve as preprocessors and specialized
interfaces for linking together other base level tools or models in a
meaningful way. This is what Ansys Sherlock does for Circuit Card Assemblies
(CCAs), with a heavy emphasis on product reliability through detailed life
In an ideal scenario, the user will have already compiled a detailed ODB++ archive containing all the relevant model information. For Sherlock, this includes .odb files for each PCB layer, the silkscreens, component lists, component locations separated by top/bottom surface, drilled locations, solder mask maps, mounting points, and test points. This would provide the most streamlined experience from a CCA design through reliability analysis, though any of these components can be imported individually.
These definitions, in combination with an extensive library of package geometries, allow Sherlock to generate a 3D model consisting of components that can be checked against accepted parts lists and material properties. The inclusion of solder mask and silkscreen layers also makes for convenient spot-checking of component location and orientation. If any of these things deviate from the expected or if basic design variation and optimization studies need to be conducted, new components can be added and existing components can be removed, exchanged, or edited entirely within Sherlock.
Figure 1: Sherlock’s 2D layer viewer and editor. Each layer can be toggled on/off, and components can be rearranged.
While a few of the available analyses depend on just the
component definitions and geometries (Part Validation, DFMEA, and CAF Failure),
the rest are in some way connected to the concept of life cycle definitions.
The overall life cycle can be organized into life phases, e.g. an operating
phase, packaging phase, transport phase, or idle phase, which can then contain
any number of unique event definitions. Sherlock provides support for vibration
events (random and harmonic), mechanical shock events, and thermal events. At
each level, these phases and events can be prescribed a total duration, cycle
count, or duty cycle relative to their parent definition. On the Life Cycle
definition itself, the total lifespan and accepted failure probability within
that lifespan are defined for the generation of final reliability metrics. Figure 1 demonstrates an example layout for a
CCA that may be part of a vehicle system containing both high cycle fatigue
thermal and vibration events, and low cycle fatigue shock events.
Figure 2: Product life cycles are broken down into life phases that contain life events. Each event is customizable through its duration, frequency, and profile.
The remaining analysis types can be divided into two categories: FEA and part specification-based. The FEA based tests function by generating a 3D model with detail and mesh criteria determined within Sherlock, which is then passed over to an Ansys Mechanical session for analysis. Sherlock provides quite a lot of customization on the pre-processing level; the menu options include different methods and resolutions for the PCB, explicit modeling of traces, and inclusion or exclusion of part leads, mechanical parts, and potting regions, among others.
Figure 3: Left shows the 3D model options, the middle shows part leads modeled, and right shows a populated board.
Each of the FEA tests, Random Vibration, Harmonic Vibration,
Mechanical Shock, and Natural Frequency, correspond to an analysis block within
Ansys Workbench. Once these simulations are completed, the results file is read
back into Sherlock, and strain values for each component are extracted and
applied to either Basquin or Coffin—Manson fatigue models as appropriate for
each included life cycle event.
Part specification tests include Component Failure Analysis for electrolytic and ceramic capacitors, Semiconductor Wearout for semiconductor devices, and CTE mismatch issues for Plated Through-Hole and solder fatigue. These analyses are much more component-specific in the sense that an electrolytic capacitor has some completely different failure modes than a semiconductor device and including them allows for a broad range of physics to be accounted for across the CCA.
The result from each type of analysis is ultimately a life
prediction for each component in terms of a failure probability curve alongside
a time to failure estimate. The curves for every component are then combined
into a life prediction for the entire CCA under one failure analysis.
Figure 4: Analysis results for Solder Fatigue including an overview for quantity of parts in each score range along with a detailed breakdown of score for each board component.
Taking it one step further, the results from each analysis are then combined into an overall life prediction for the CCA that encompasses all the defined life events. From Figure 5, we can see that the life prediction for this CCA does not quite meet its 5-year requirement, and that the most troublesome analyses are Solder Fatigue and PTH Fatigue. Since Sherlock makes it easy to identify these as problem areas, we could then iterate on this design by reexamining the severity or frequency of applied thermal cycles or adjusting some of the board material choices to minimize CTE mismatch.
Figure 5: Combined life predictions for all failure analyses and life events.
Sherlock’s convenience for defining
life cycle phases and events, alongside the wide variety of component
definitions and failure analyses available, really cement Sherlock’s role as a
comprehensive electronics reliability tool. As in most analyses, the quality of
the results is still dependent on the quality of the input, but all the checks
and cross-validations for components vs life events that come along with
Sherlock’s preprocessing toolset really assist with this, too.
For those who are not already familiar with it, Discovery
Live is a rapid design tool that shares the Discovery SpaceClaim environment.
It is capable of near real-time simulation of basic structural, modal, fluid,
electronic, and thermal problems. This is done through leveraging the computational
power of a dedicated GPU, though because of the required speed it will
necessarily have somewhat less fidelity than the corresponding full Ansys
analyses. Even so, the ability to immediately see the effects of modifying,
adding, or rearranging geometry through SpaceClaim’s operations provides a
tremendous value to designers.
One of the most interesting features within Discovery Live
is the ability to perform Topology Optimization for reducing the quantity of
material in a design while maintaining optimal stiffness for a designated
loading condition. This can be particularly appealing given the rapid adoption
of 3D printing and other additive manufacturing techniques where reducing the
total material used saves both time and material cost. These also allow the
production of complex organic shapes that were not always feasible with more traditional
techniques like milling.
With these things in mind, we have recently received
requests to demonstrate Discovery Live’s capabilities and provide some training
in its use, especially for topology optimization. Given that Discovery Live is
amazingly straightforward in its application, this also seems like an ideal
topic to expand on in blog form alongside our general Discovery Live workshops!
For this example, we have chosen to work with a generic
“engine mount” geometry that was saved in .stp format. The overall dimensions
are about 10 cm wide x 5 cm tall x 5 cm deep, and we assume it is made out of
stainless steel (though this is not terribly important for this demonstration).
Figure 1: Starting engine mount geometry with fixed supports and a defined load.
The three bolt holes around the perimeter are fixed in
position, as if they were firmly clamped to a surface, while a total load of
9069 N (-9000 N in X, 1000 N in Y, and 500 N in Z) is applied to the cylindrical
surfaces on the front. From here, we simply tell Discovery Live that we would
like to add a topology optimization calculation onto our structural analysis.
This opens up the ability to specify a couple more options: the way we define
how much material to remove and the amount of material around boundary
conditions to preserve. For removing material, we can choose to either reduce
the total volume by a percent of the original or to remove material until we
reach a specific model volume. For the area around boundary conditions, this is
an “inflation” length measured as a normal distance from these surfaces, easily
visualizable when highlighting the condition on the solution tree.
Figure 2: Inflation zone shown around each fixed support and load surface.
Since I have already planned out what kind of comparisons I
want to make in this analysis, I chose to set the final model volume to 30 cm3.
After hitting the simulate button, we get to watch the optimization happen
alongside a rough structural analysis. By default, we are provided with a
result chart showing the model’s volume, which pretty quickly converges on our
target volume. As with any analysis, the duration of this process is fairly
sensitive to the fidelity specified, but with default settings this took all of
7 minutes and 50 seconds to complete on my desktop with a Quadro K4000.
Figure 3: Mid-optimization on the top, post-optimization on the bottom.
Once optimization is complete, there are several more operations that become available. In order to gain access to the optimized structure, we need to convert it into a model body. Both options for this result in faceted bodies with the click of a button located in the solution tree; the difference is just that the second has also had a smoothing operation applied to it. One or the other may be preferable, depending on your application.
Figure 4: Converting results to faceted geometry
Figure 5: Faceted body post-optimization
Figure 6: Smoothed faceted body post-optimization
Though some rough stress calculations were made throughout
the optimization process, the next step is typically a validation. Discovery
Live makes this as a simple procedure as right-clicking on the optimized result
in the solution tree and selecting the “Create Validation Solution” button.
This essentially copies over the newly generated geometry into a new structural
analysis while preserving the previously applied supports and loads. This
allows for finer control over the fidelity of our validation, but still a very
fast confirmation of our results. Using maximum fidelity on our faceted body,
we find that the resulting maximum stress is about 360 MPa as compared to our unoptimized
structure’s stress of 267 MPa, though of course our new material volume is less
than half the original.
Figure 7: Optimized structure validation. Example surfaces that are untouched by optimization are boxed.
It may be that our final stress value is higher than what we
find acceptable. At this point, it is important to note one of the limitations
in version 2019R3: Discovery Live can only remove material from the original
geometry, it does not add. What this means is that any surfaces remaining
unchanged throughout the process are important in maintaining structural
integrity for the specified load. So, if we really want to optimize our
structure, we should start with additional material in these regions to allow
for more optimization flexibility.
In this case, we can go back to our original engine mount
model in Discovery Live and use the integrated SpaceClaim tools to thicken our
backplate and expand the fillets around the load surfaces.
Figure 8: Modified engine mount geometry with a thicker backplate and larger fillets.
We can then run back through the same analysis, specifying
the same target volume, to improve the performance of our final component.
Indeed, we find that after optimizing back down to a material volume of 30 cm3,
our new maximum stress has been decreased to 256 MPa. Keep in mind that this is
very doable within Discovery Live, as the entire modification and simulation
process can be done in <10 minutes for this model.
Figure 9: Validated results from the modified geometry post-optimization.
Of course, once a promising solution has been attained in Discovery Live, we should then export the model to run a more thorough analysis of in Ansys Mechanical, but hopefully, this provides a useful example of how to leverage this amazing tool!
One final comment is that while this example was performed in the 2019R3 version, 2020R1 has expanded Discovery Live’s optimization capability somewhat. Instead of only being allowed to specify a target volume or percent reduction, you can choose to allow a specified increase in structure compliance while minimizing the volume. In addition to this, there are a couple more knobs to turn for better control over the manufacturability of the result, such as specifying the maximum thickness of any region and preventing any internal overhangs in a specified direction. It is now also possible to link topology optimization to a general-purpose modal analysis, either on its own or coupled to a structural analysis. These continued improvements are great news for users, and we hope that even more features continue to roll out.
The role of Ansys Electronics Desktop Icepak (hereafter referred to as Icepak, not to be confused with Classic Icepak) is in an interesting place. On the back end, it is a tremendously capable CFD solver through the use of the Ansys Fluent code. On the front end, it is an all-in-one pre and post processor that is streamlined for electronics thermal management, including the explicit simulation and effects of fluid convection. In this regard, Icepak can be thought of as a system level Multiphysics simulation tool.
One of the advantages of Icepak is in its interface consistency with the rest of the Electronic Desktop (EDT) products. This not only results in a slick modern appearance but also provides a very familiar environment for the electrical engineers and designers who typically use the other EDT tools. While they may not already be intimately familiar with the physics and setup process for CFD/thermal simulations, being able to follow a very similar workflow certainly lowers the barrier to entry for accessing useful results. Even if complete adoption by these users is not practical, this same environment can serve as a happy medium for collaboration with thermal and fluids experts.
Figure 1: AEDT Icepak interface. The same ribbon menus, project manager, history tree, and display window as other EDT products.
So, beyond these generalities, what does Icepak actually
offer for an optimized user experience over other tools, and what kinds of
problems/applications are best suited for it?
The first thing that comes to mind for both of these
questions is a PCB with attached components. In a real-world environment,
anyone that has looked at the inside of a computer is likely familiar with
motherboards covered with all kinds of little chips and capacitors and often
dominated by a CPU mounted with a heatsink and fan. In most cases, this motherboard
is enclosed within some kind of box (a computer case) with vents/filters/fans
on at least some of the sides to facilitate controlled airflow. This is an
ideal scenario for Icepak. The geometries of the board and its components are typically
well represented by rectangular prisms and cylinders, and the thermal
management of the system is strongly related to the physics of conjugate heat
transfer. For the case geometry, it may be more convenient to import this from
a more comprehensive modeler like SpaceClaim and then take advantage of the
tools built into Icepak to quickly process the important features.
Figure 2: A computer case with motherboard imported from SpaceClaim. The front and back have vents/fans while the side has a rectangular patterned grille.
For a CAD model like the one above, we may want to include
some additional items like heatsinks, fan models, or simple PCB components.
Icepak’s geometry tools include some very convenient parameterized functions
for quickly constructing and positioning fans and heatsinks, in addition to the
basic ability to create and manipulate simple volumes. There are also routines
for extracting openings on surface, such as the rectangular vent arrays on the
front and back as well as the patterned grille on the side. So, not only can
you import detailed CAD from external sources, you can mix, match, and simplify
it with Icepak’s geometry, which streamlines the entire design and setup
process. For an experienced user, the above model can be prepared for a basic
simulation within just a matter of minutes. The resulting configuration with an
added heatsink, some RAM, and boundary conditions, could look something like
Figure 3: The model from Figure 2 after Icepak processing. Boundary conditions for the fans, vents, and grille have been defined. Icepak primitives have also been added in the form of a heatsink and RAM modules.
Monitor points can then assigned to surfaces or bodies as
desired; chances are that for a simulation like this, temperature within the
CPU is the most important. Additional temperature points for each RAM module or
flow measurements for the fans and openings can also be defined. These points
can all be tracked as the simulation proceeds to ensure that convergence is
Figure 4: Monitoring chosen solution variables to ensure convergence.
For this simple system containing a 20 W CPU and 8 RAM
modules at 2 W each, quite a few of our components are toasty and potentially
problematic from a thermal standpoint.
Figure 5: Post-processing with Icepak. Temperature contours are overlaid with flow velocities to better understand the behavior of the system.
With the power of a simulation environment in Icepak at our
fingertips, we can now play around with our design parameters to improve the thermal
management of this system! Want to see what happens when you block the outlet
vents? Easy, select and delete them! Want to use a more powerful fan or try a
new material for the motherboard or heatsink? Just edit their properties in the
history tree. Want to spin around the board or try changing the number of fins
on the heatsink? Also straightforward, although you will have to remesh the
model. While these are the kinds of things that are certainly possible in other
tools, they are exceptionally easy to do within an all-in-one interface like
The physics involved in this example are pretty standard:
solid body conduction with conjugate heat transfer to a turbulent K-Omega fluid
model. Where Icepak really shines is its ability to integrate with the other
tools in the EDT environment. While we assumed that the motherboard was nothing
more than a solid chunk of FR-4, this board could have been designed and
simulated in detail with another tool like HFSS. The board, along with all of
the power losses calculated during the HFSS analysis, could have then been
directly imported into the Icepak project. This would allow for each layer to
be modeled with its own spatially varying thermal properties according to trace
locations as well as a very accurate spatial mapping of heat generation.
This is not at all to say that Icepak is limited to these
kinds of PCB and CCA examples. These just often tend to be convenient to think
about and relatively easy to geometrically represent. Using Fluent as the
solver provides a lot of flexibility, and there are many more classes of
problems that could be benefit from Icepak. On the low frequency side, electric
motors are a good example of a problem where electronic and thermal behavior
are intertwined. As voltage is applied to the windings, currents are induced
and heat is generated. For larger motors, these currents, and consequently the associated
thermal losses, can be significant. Maxwell is used to model the electronic
side for these types of problems, where the results can then be easily brought
into an Icepak simulation. I have gone through just such an example
rotor/stator/winding motor assembly model in Maxwell, where I then copied everything
into an Iecpak project to simulate the resulting steady temperature profile in
a box of naturally convecting air.
Figure 6: An example half-motor that was solved in Maxwell as a magnetostatic problem and then copied over to Icepak for thermal analysis.
If it is found that better thermal management is needed,
then extra features could then be added on the Icepak side as desired, such as
a dedicated heatsink or external fan. Only the components with loads mapped
over from Maxwell need to remain unmodified.
On the high frequency side, you may care about the
performance of an antenna. HFSS can be used for the electromagnetic side, while
Icepak can once again be brought in to analyze the thermal behavior. For high powered
antenna, some components could very easily get hot enough for the material
properties to appreciably change and for thermal radiation to become a dominant
mode of heat transport. A 2-way automatic Icepak coupling is an excellent way
to model this. Thermal modifiers may be defined for material properties in
HFSS, and radiation is a supported physics model in Icepak. HFSS and Icepak can
then be set up to alternately solve and automatically feed each other new loads
and boundary conditions until a converged result is attained.
What all of this really comes down to is the question: how
easy is it for the user to set up a model that will produce the information
they need? For these kinds of electronics questions, I believe the answer for
Icepak is “extraordinarily easy”. While functional on its own merit, Icepak
really shines when it comes to the ease of coupling thermal management analysis
with the EM family of tools.