– This warning appears because the separation (think of it as a tolerance setting for the mesher) distance is larger than 10% of the smallest feature in the model.
– When there are objects smaller than the mesher tolerance, those objects will not be meshed correctly.
– However, note that the separation setting is a useful tool designed to avoid unnecessary meshes due to inadvertent misalignments in the model (without modifying the geometry).
– Look for the name of the object featured in the warning and its dimension.
– Why should you model an object that's 0.1 mm in thickness? Is it likely to improve the accuracy of your results?
– Do not accept the suggested change to the separation settings.
8.12. Modification 5: A Simplification Based on Magnitudes of Resist-ances...
• From the name of the object, one can infer that the warning is regarding an air gap under one of the components, which is modeled as a thick plate.
• There is a reason for not using contact resistance type plate to model the Airgap.
– Two thin objects cannot overlap. If the Airgap was modeled as contact resistance plate, the under-lying mask may not be meshed in the region common to the Mask and the Airgap.
→ What is the Mask thickness and conductivity?
→ What is the Airgap thickness and conductivity?
→ The purpose for modeling these two objects is to capture their insulating effects.
→ How does the resistance (thickness/conductivity) of the mask compare to that of the Airgap?
→ Does the mask contribute significantly to the overall (sum) of the two resistances?
→ Can you justify suppressing the mask under the air gap by making the Airgap a contact resistance plate?
→ When you make the Airgap a contact resistance plate, make sure that the Effective thickness is the same (1e-4 m).
→ Also make sure the Airgap has higher priority over the Mask object.
• You can do this by editing the plate object and changing the Priority setting under the Info tab. (Larger priority number means higher priority. Objects with higher priority are listed lower in the Model manager window).
• Generate the mesh again.
• This time you will see another separation warning about the AL-spreader. Again, do not accept the changes.
8.13. Modification 6: A Classic Case for Thin Conducting Plate...
• Since a contact resistance plate will not model the in-plane spreading of heat, we can't use it here.
Thin conducting plate models conduction in the normal as well as the planar direction. At the same time the thin conducting plate will not generate slender cells.
The decrease in thickness due to a thin plate approximation of the Al-spreader is negligible.
• Generate the mesh one more time. You will see the separation warning again - this time about the die objects which are 0.0004 m.
– These objects are power generating components, which are already thin conducting plates. The warning is about the width of the packages.
– The surface area of the dies is a critical parameter affecting the temperature prediction for the component. This cannot be simplified.
– Hence accept the suggested change in separation setting. The resultant mesh count will be signific-antly less than what we got without any changes to the given model.
Note
It is also possible to use a separation distance larger than the recommended 10% value.
Values of up to 50% (of the smallest dimension) may be used in cases where reducing the mesh count is critical.
• You will now get a separation warning about the tabs. We cannot change the geometry of the tabs, so accept the suggested change in separation settings again.
• Here are some suggested qualities of meshes:
– The size of the first cells from critical heat dissipating surfaces should be less than 1 mm for a 1st cut analysis.
→ View mesh cut plane on the wall of the enclosure object, the PCB and the critical heat generating components to see if you are fulfilling the above requirement.
→ Use the Object params control to request mesh refinement near all the important surfaces mentioned above.
– Generate the mesh to see if your request is being honored.
• Finally, a comparison...
For comparison purposes, deactivate the Mesh assemblies separately option in the Mesh control panel and generate the mesh. The difference between the mesh with this check button active and inactive is the effect of non-conformal meshing.
• STOP: Solution and post processing are beyond the scope of this exercise. Please compare the suggested approach with the approach you were attempting during the initial 15 minute period of this tutorial.
8.14. Conclusion
A model with room for improvement is provided. Using approximate object choices and meshing strategies, the model and the mesh were improved. The approach delineated in this exercise can help reduce significant run time without compromising the physics being modeled.
9.1. Introduction
This tutorial demonstrates how to define trials, run parametric solutions, and post-process the results.
Often, there is a need to calculate the loss coefficient of grilles that have certain hole patterns. The purpose of the problem is to determine the minor loss coefficient of a grille that has hexagonal holes.
In this tutorial you will learn how to:
• Define a parameter to optimize the design.
• Define trials.
• Define primary and compound functions that you want to report.
• Calculate parametric solutions.
• Report and plot parametric results.
9.2. Prerequisites
This tutorial assumes that you are familiar with the menu structure in ANSYS Icepak and that you have solved or read the tutorial "Finned Heat Sink". Some steps in the setup and solution procedure will not be shown explicitly.
9.3. Problem Description
The model includes a cabinet that is 160 mm in length with inlet and outlet openings at the two ends (with cross sectional area of 7.363 mm x 12.7 mm), and four symmetry walls at the other sides. The model also includes a part of the hexa-grille placed at the center of the channel in the streamwise dir-ection, as shown in Figure 9.1 (p. 144). The grille has one full hexagonal hole at the center and four quarter hexagonal holes placed around it. This pattern was selected because it forms a periodic region and is sufficient to calculate the loss coefficient. The solution obtained from this run can be replicated to form the solution for the entire domain.