Liner segmentation
 

Hi,


I need to collect the heat transfer coefficient and the temperature data (near-wall cell data) on different segments of liner to conduct FEM simulation in another program. The question is how do I discrete my liner and collect these data?
One solution that may possible is to divide the wall liner into the multiple boundaries and check the bound_wall#.out in the output results. Happy to hear the exact solution.


Thanks,

Hello Hamed,

You can direct CONVERGE to write out boundary/near-wall heat transfer data (htc, near-wall temperatures, heat flux, wall stress etc at different XYZ locations for cells along the walls) to a transfer.out file. This can be enabled through CONVERGE Studio : Output/Post-Processing > Output files > Heat transfer output > Generate local heat transfer data (or transfer_flag = 1 in inputs.in). The output frequency is controlled by twrite_transfer in inputs.in and you can also direct CONVERGE to only write out data for cells adjacent to a list of boundaries using a transfer.in file. After the simulation is done, you can make use of our HTC mapping utility to cycle-average the transfer.out data on a FEA mesh and feed it to your FEA tool.

For more details on the above approach, please refer our manual or the Heat Transfer Mapping advanced training slides. The HTC mapping utility can be downloaded from our website and includes a tutorial for an IC engine model.

If your Piston is attached to the liner, which is often the case unless you're running two-stroke with sealing, you cannot divide the wall liner into multiple boundaries. This will create vertices/triangles along the liner which will intersect with the piston as it moves up.

Hope this helps.

Sincerely,

Hello Kislaya,

Thank your for your response.

I mapped the fluid temperature data from the cells near the liner to a similar liner surface having FEA mesh, but something went wrong (see attached image, just for intake and compression simulation). I think the piston shape also mapped to the liner because the crevice (top land) added to the piston bowl, which is order of 0.4 mm wide, has insufficient resolution mesh size. That is, the cells near to the wall are the same for both liner and piston. The question is: How can this flaw affect the entire simulation (piston motion affect the liner by its wall temperature or heat release during combustion event)?

Best regards,

Hello Hamed,

Your selection of wall boundary temperatures will influence the near-wall temperatures for the fluid. What is the boundary temperature you have provided for your piston top land surface? If you have just one cell between the liner and the top land, and if the piston land boundary temperatures are hotter, then the poorly refined cells across the crevice will heat up due to wall heat transfer, leading to elevated temperatures for the fluid that are then mapped onto the liner surface, since these fluid cells are shared with the piston land surface. I would recommend refining your crevice volume, to have 2-4 cells across the crevice gap.

Such a behaviour might be more prominent since you only simulated intake-compression cycle. If you're mapping HTC results to an FEA solver, you'd have to do it for the complete engine cycle. Also, please confirm your HTC mapping process, using our utility, was as per our recommendations from our tutorial/quick setup guide (available on our Downloads page).

Sincerely,

Thank you again.

There are other questions:


1- Assuming that a crevice volume has a width in the order of 0.5 mm which requires a fixed embedding with scale of 4 (base grid = 2.8 mm) to capture 2-4 cells between gap. Regardless of hot engine conditions that make this worse, y+ cannot fall into the range of 30-100 (standard wall function), even with using release and boundary AMR.

2- How to manage transfer.out file such that different boundaries have different files? If there is one file for HTC mapping, that is too big, the separation of different walls is just a waste of time.

Best regards,
Hamed

Hello Hamed,

1. There is no upper limit to your selection for embed scale for fixed embedding and boundary (y+ target) AMR. If you use an appropriate value, you will be able to make your cell sizes fine enough for the y+ values to fall into the 30-100 range. In my experience, CONVERGE HT results have been robust and accurate for y-plus values higher than 100. So we don't often need such fine of a grid.
Increase in embed scale will increase your cell count, but you have to do this if you want to bring your yplus values down. But you can also consider using a moving inlaid mesh (available in v3.1) for the crevice region as a more optimal meshing approach. User generated grid can be constructed for the crevice volume with conformed cell sizes refined only in the direction normal to the wall. This will enable lower cell count compared to our autonomous Cartesian grid which is always refined in all 3 directions.

2. Heat transfer data will be written out to a single transfer.out file. You can only provide a limited list of boundaries you need data for using the transfer.in file. However, with the HTC mapping utility, you can map heat transfer data to individual surfaces independently. Also, when mapping on the full surface, HTC mapped data (*.map files created) lists out boundary ID for each XYZ location. You should be able to easily separate/sort data points by Bound_ID using simple scripts or Excel.

Hope this helps.

Sincerely,