peteroznewman
Subscriber

Wenlong outlined a very nice decision tree.  The process you describe sounds like Thermoforming. I understand the plastic sheet will be heated in an oven to a uniform temperature, then placed in a mould and pressed into shape and allowed to cool so that it keeps the shape when removed from the mould.

The heat flow between the plastic and the mould can affect the thickness distribution as the pressing occurs. If the mould is wood, then the heat transfer to the mould while pressing is slow, but if the mould is aluminum, then the heat transfer during pressing could be large, reducing the temperature of the plastic in the contact area and affecting where the thinning occurs.

• What material is the mould made out of?

• What is the temperature of the mould when the plastic is inserted?

• What is the temperature of the plastic when inserted?

• What is the air temperature around the mould?

• How many seconds does it take to move the plastic from the oven to the mould?

• How many seconds does it take the mould to close and press the shape (not counting the cooling time)?

The type of analysis depends on the answers to these questions.

I have the free Student license of ANSYS 2020 R1 and the Coupled Field Statics is the first analysis in the Toolbox.

A Statics model can have a large amount of deformation in the parts which is enabled under Analysis Settings by turning on Large Deflection.

Looking at your image, I see a single layer of elements through the thickness of the plate. I can't tell if you modeled the plastic sheet as a surface in CAD and meshed it with shell elements and assigned a thickness, or if you modeled the plastic sheet as a solid in CAD.  If you have solid elements, you will require around 8 elements through the thickness of the sheet. This will require a lot of nodes and you might exceed the Student node limit of 32,000.  I recommend you create a surface in CAD and mesh with shell elements. The thickness of the shell can be tracked during the solution and the thickness distribution easily shown in the results. This model will also take less time to solve.