Frequently Asked Questions
- Who can purchase the manual?
Anyone can purchase online access to the manual - individuals, companies, or other organizations.
- What does the manual cost?
A one-year license to access the manual electronically costs USD $190.
- Do you sell multi-user licenses?
The $190 one-year license is for a single user. Institutions such as Universities should contact us for pricing information for multiple concurrent user access.
- Is there a printed edition of the manual?
The manual is only available online at this time. An advantage of this arrangement is that updates can be made to the manual without the need for reprinting.
- Does the manual also teach how to create 2D meshes? or is it only 3D?
A 2D mesh is a sector mesh of 0.5 deg of only one cell - so one can create a 2D mesh by following the instructions for making a sector meshfound in Chapter 2.
- What is the version of ICEM CFD used in this manual?
The manual uses ICEM CFD 12.1.
- When creating a mesh using the manual, we encounter a problem with the itape17 file of the final mesh. The error message says:
WARNING: cell 55 in region 1 is not convex:
WARNING: 470 non-convex cells at cycle 0, crank = 0.00 -- see otape12
How can we solve this problem?
There is a well known problem within ICEM CFD when exporting a sector mesh. It is a tolerance issue in the valve axis nodes. Occasionally the x and y coordinates of the axis nodes are not exactly zero at some decimal digit after the decimal point (for example instead of of being 0.00000000 it could be 0.00000001), and the kiva soubroutine that controls the convexity marks those axis cells as non-convex.
Usually the Kiva calculation runs without any problem. But you can easily insert a loop in Kiva that transforms the x and y nodes under a certain threshold into exactly zero with the kiva required tolerance.
- When exporting the ICEM mesh in KIVA format, we encounter a problem where the mesh looks fine in ICEM (both mesh and pre-mesh), but after converted to itape17 and viewed in Ensight, several cells are "missing", as if they are taken off from their resided block. The block associates to a very thin layer between the needle and seat of a fuel injector. Since these cells are missing, KIVA generates error when setting up Boundary Conditions.
Try starting a new blank project with only your geometry file (.tin) and your blocking file (.blk). Then use the Run-Check-fix tool for the blocks and try generating the KIVA input file again.
Thanks! We found that the IJK order of two local blocks were messed up. Therefore, cell connectivity was wrong in itape17. Now the problem is solved.
- In Chapter 4.2, step 10 directs the user to "move and associate vertices and edges accordingly like in step 7." But this step doesn't indicate the vertices, edges and curves in detail. We are unable to duplicate figure 4.16. Can you help?
Actually is very simple, at the lower K-plane of the bowl, you need to associate the bowl border edges to the bowl curve (There is only one curve in the XY plane).
Checking the remaining part of the chapter should make this step clearer as well.
- How can we create an external program to change flags around the valve?
For the GDI grid in the last chapter of the manual, the content of the program is explained in detail and is grid dependent - so you have to create it based on the number of valve stem nodes.
- It is recommended in the manual that cells under mesh quality of 0.2 should be avoided. After we modified all the blocks, there are still several cells under -0.2, with the total number less than 5. Is it OK for calculation?
Regarding the quality of the grid, it really depends on the level. For example even if only one cell is below a certain value the calculation will not work. I try to work when possible with quality greater than 0 to avoid computational errors. But the GDI grid does have some cells between 0 and -0.2.
- For the sector mesh, there's always something wrong with the boundary conditions of the cells around the axis. How can we fix it?
You can try to associate faces with i=1 to the axis to fix it.
- When I generate a sector mesh and set up a case, there seems to be a node near the axis not moving with the piston movement. How can we fix it?
This is the block index problem. Typically, the block index with lower k-plane should be smaller than that with higher k-plane. When the block index is out of order, we can rearrange it using Renumber Blocks tool in Edit Block. But we can also avoid this by splitting blocks along x+ direction, y+ direction and then z+ direction in order from the beginning.
- For the engine mesh there's always something wrong with the boundary conditions of the cells around the valves. How can we fix it?
It needs a shared wall operation between the squish block and intake/exhaust port block.