Abaqus: Heating of Ti plate using DFLUX subroutine:

[Amit2]

Hi Friends, I am a beginner in the Abaqus and trying to understand the impact of different heat source on the metal plate. For the current case, I am using Gaussian distribution heat source to simulate the melting in the Ti plate. So my thermal model uses temperature-dependent specific heat, conductivity, latent heat and density. Also, the sample size of 0.8X0.8X0.4mm with the mesh size of 0.8/200. To give desired surface heat flux I am using DFLUX subroutine.

My units look consistent, I am using the metric standard where my conductivity varies from 4 to 40 W/mK wrt temp, Density 4400Kg/m3, specific heat order of 750 below MP and 950 J/Kg-K after MP. Attributes consider 0 K temperature and metric Boltzmann constant 5.67*10-8.

My model considers 70-watt Gaussian power with an effective diameter of 125micro-meter for 400 microseconds. Therefore I cannot change my maximum time increment beyond 400us. I am using DC3D8: An 8-node linear heat transfer brick.
Now the problem I am facing is the following:

1. The surface of model is exposed to a very high temperature but the temperature is not conducting to the next layer of nodes, please check the image. I read some journal where they reported that the depth of melting goes up to 120-200micron. My simulation shows just 2 microns. I have to know where I am doing wrong.

2. Simulation is not completing, either it says time step is too big (for fixed time step) or it says Too many attempts made for this increment (Automatic steps). I used other time steps and mesh size it solves but I don't know what I am doing and why I am doing? If you can explain or give some reference from where I can read and learn how to adjust the step size?.

 

Welding, Bonding & Fastener engineering is probably not the best place to ask this.

"Everyone is entitled to their own opinions, but they are not entitled to their own facts."

 

[DrZoidberWoop]

Amit2,

I used Abaqus extensively when I was at University, and I encountered many technical issues over various topics. The best way to troubleshoot your model is to find an extremely simplified version of your problem that satisfactorily converges,and extend the complexity incrementally, unless you are conducting something that doesn't require convergence like an element stability problem, XFEM cracking, or element enrichment/failure. There are a great number of variables which may be the cause of your nonconvergence, many of which I'm totally unfamiliar with. I imagine that you have already studied this person's tutorial?

https://www.youtube.com/watch?v=3CoErpKTpV4

"ABAQUS Tutorial | Welding Analysis with DFLUX subroutine | Coupled Temp-Displacement | 16-25"

He used a series of coupled temperature displacement steps with C3D8T elements. It has always been very difficult to find answers to problems like this because you are interested in a very specific field which requires advanced theoretics.

"I used other time steps and mesh size it solves but I don't know what I am doing and why I am doing?" This is not an acceptable answer for a boss or evaluation committee. Please be very careful in your reporting.

I'd also like to say, if you do not fully understand the theoretics (elasticity, plasticity, PDEs within heat transfer, etc.), and their implications/limitations within the finite element analysis framework, then you are only using ABAQUS to create pretty pictures, which is deceptive at best. I used to see this all the time at academic conferences, and it puts you in a very vulnerable position if you can't satisfactorily explain even a tiny portion of your model. I don't mean to sound grim, and I wish you the best of luck, but I've seen too much fraudulent "research" over the years disguised behind a facade of beautiful contour plots.