Learning Forum

[Ansuman_Sahu]

I am performing a transient thermal analysis of fused deposition modelling process using ANSYS APDL (Ansys 2020 R2 - Student version). I am trying to simulate the deposition of a single extruded strand using element birth and death technique. In each load step one element is activated. The initial temperature of the activated element (in the current load step) is to be the nozzle temperature. The initial temperatures of the previously activated elements is obtained from the results (temperature field) of the previous load step.

1. The steps adopted in the model are as follows:
2. Creating & meshing the 3D model (of a single extruded strand of length=50mm)
3. Deactivating all elements (using EKILL)
4. Setting up the solution
5. Activating one element
6. Specifying initial condition on nodes
7. Specifying boundary conditions (Convection load on exterior node)
8. Specifying the necessary load step options and solving.
9. Repeating steps 4 to 7 for each activated element (using *DO loop).

I am facing the following issues:

1. The analysis runs properly only for the first load step, in that the decrease in temperature of the element is in accordance with the convective loading applied on it. But from the second and subsequent load steps onwards, the temperature distribution obtained is not proper. When the second element is activated, the face of the element exposed to ambient air has a constant temperature = nozzle temperature even though convective load is applied on this element. Element numbers 3 and beyond, remain entirely at the constant temperature = nozzle temperature. Note: If however, each load step is run in a separate file (with all other parameters constant and by reading in the solution of the previous load step as the initial condition), the results of each load step are in accordance with the boundary conditions and loads assigned to them. I am unable to identify what's causing these improper nodal solutions when solving all the load steps in a single file. Am I missing out on any crucial step in my code? Why is there a difference in nodal solution between these two approaches (solving all load steps in one file v/s solving each load step in a separate file?
2. When assigning initial conditions, the temperature distribution from the previous load steps can be read (using LDREAD) for the previously activated elements. However, I am unable to separately provide the initial temperature to the element activated in the current load step. The following warning message gets generated : Initial Conditions not allowed after first Load Step. Currently I have managed to avoid this issue by assigning a uniform temperature (equal to the nozzle temperature) at the beginning of solution for all the nodes (of activated as well as deactivated elements), but I am not sure of it. Is there a proper way to separately provide the initial condition for element activated in the current step?
3. Also, is LDREAD command the proper method for using the results of the previous load step as the initial condition for the current step? 

Since I am new to ANSYS, any suggestions w.r.t these two issues and help in identifying the mistakes in my code/approach will be greatly appreciated.

The relevant screenshots from APDL as well as the code have been attached for reference.

[Dave Looman]

Your description of the issue is correct. It is difficult to make the initial temperature of an element what you want when it is activated. When elements are inactive (dead) they are still there in the finite element model, they just have watered down material properties. As a result, they can change temperature through conduction from the adjacent active elements. One approach to giving them a desired initial temperature is to constrain their nodes to that temperature while they are inactive and then delete those constraints when they are activated. This method has the drawback that the inactive elements share nodes with the adjacent active elements and those nodes can't be constrained. A thin layer of inactive elements at the interface might help this. Another method is to apply heat to the elements when they are activated to bring them up to the desired temperature. n

[Ansuman_Sahu]

Thank you very much for these suggestions. I am sorry for replying late. nI tried the first suggestion of constraining the nodes of the inactive elements at the initial temperature, and got a reasonable spatial temperature distribution and thermal histories. However, the drawback, which you have mentioned was quite clear from the contour plots. Although I have managed to resolve it to some extent by using a finer mesh, but the computation time is certainly quite high now and also because I am using a student version, there's the limitation on the number of nodes/elements. nI will definitely work on the other two approaches which you have suggested. n