Solidification with a Thermal Gradient in L-PBF (Equiaxed+Columnar)

dendritic solidification, eutectics, peritectics,....
Post Reply
Moritz
Posts: 40
Joined: Fri Aug 06, 2021 11:13 am
anti_bot: 333

Solidification with a Thermal Gradient in L-PBF (Equiaxed+Columnar)

Post by Moritz » Wed Oct 13, 2021 11:59 am

Hi all,

I am working with the author of the topic before (viewtopic.php?f=9&t=746) on a simulation of the solidification after the LPBF process.

We initially tried different approaches to achieve a combined columnary and equiaxed solidification. When we applied a thermal profile on the T002_AlCu example, which is resembling the thermal conditions during the scanning motion of the laser in LPBF, we were initially happy with the results. Besides a planar solidification from the top and the details of the solidification starting in the melt pool, this is what we want to see with our alloy.

Image

I changed the driving file according to our needs (database, concentrations etc.) but the solidification changed massively.

Image

It is my opinion, that the solidification behaviour on this scale is mostly influenced by the thermal conditions. Of course, I expected some changes but espacially the missing columnary growth is confusing to me.

Do you have any suggestions, what paramters can also be changed to influence the solidification?

I am thankful for your suggestions.

Best regards,

Moritz

Bernd
Posts: 1504
Joined: Mon Jun 23, 2008 9:29 pm

Re: Solidification with a Thermal Gradient in L-PBF (Equiaxed+Columnar)

Post by Bernd » Wed Oct 13, 2021 12:31 pm

Dear Moritz,

I am wondering how you adapted the T002_AlCu_Equiaxed_dri example to LPBF conditions. For additive applications it does not make much sense to use latent heat in this form ("DTA-approximation"), and you need to have a strong temperature gradient and cooling rate. Looking at your results suggests that you did not have a considerable temperature gradient. Which thermal boundary conditions did you actually apply?

In my opinion, the most promising thermal boundary conditions for LPBF are:

a) using constant temperature gradient and cooling rate (most simple)

b) using time-dependent temperature "profiles_from_file", which come from process-scale simulations or analytical approximations

c) including latent heat by coupling to an internal 1d-temperature field ("1d_temp", my current topic of development)

Bernd

Moritz
Posts: 40
Joined: Fri Aug 06, 2021 11:13 am
anti_bot: 333

Re: Solidification with a Thermal Gradient in L-PBF (Equiaxed+Columnar)

Post by Moritz » Thu Oct 14, 2021 9:45 am

Hi Bernd,

thanks for your reply. I obviously left some important information out.

We did not use latent heat in our approach. The latent heat effects are included in the temperature profile, which comes from a FEM-model. We tested the application of thermal data on the T002 example, but then changed it according to our needs.

I attach the driving file and the temperature profile for a better understanding.

Best regards,

Moritz
Attachments
T021_Temperature_Profiles.txt
(187 Bytes) Downloaded 148 times
AM100.dri
(19.93 KiB) Downloaded 169 times

Bernd
Posts: 1504
Joined: Mon Jun 23, 2008 9:29 pm

Re: Solidification with a Thermal Gradient in L-PBF (Equiaxed+Columnar)

Post by Bernd » Thu Oct 14, 2021 2:41 pm

Hi Moritz,

Please check the .temp output and its time development and whether the temperature boundary condition works like you expect. I see two reasons why it could be wrong:

1.) The tabulated input which you are using is in both dimensions (position in z-direction and time) smaller than your domain / simulation time. This means that extrapolation is required in both dimensions

2.) The z-position is typically specified in decreasing order. I am not 100% sure whether reversed order could not lead to additional problems

Bernd

Moritz
Posts: 40
Joined: Fri Aug 06, 2021 11:13 am
anti_bot: 333

Re: Solidification with a Thermal Gradient in L-PBF (Equiaxed+Columnar)

Post by Moritz » Mon Oct 25, 2021 2:30 pm

Hi Bernd,

thanks for your reply. This was actually a good idea, to check the inputs again. Somehow they worked for one simulation, but transfered to the other one caused problems.

We have now a simulation, where we considered your suggestions and the simulaion is starting to look like we expect it to.

Image

Thank you very much!

Best regards,

Moritz

Post Reply