Learning Forum

[romain.gaspar]

Hello everyone,

I am trying to perform a simulation in Polyflow using a viscoelastic material model, but I am obtaining results that are completely different from my Newtonian reference case.

Reference Newtonian case

I first set up an isothermal Newtonian filament stretching simulation (viscosity = 80 pa.s) and a strong stretching force, which runs correctly. The fiber attenuates as expected and after about 0.1 s a long fiber is already formed.

Viscoelastic case

I then duplicated the exact same setup and only changed the material model to a Differential Viscoelastic Isothermal Flow model (I tested both Maxwell and Oldroyd-B). 

For simplicity:

• Constant viscosityof 80 pa.s
• Same geometry, mesh, boundary conditions and operating conditions as the Newtonian case
• Activated "Couple computation of viscoelastic stresses"
• Very small relaxation time

In particular, I set the relaxation time to λ = 10⁻¹⁰ s, expecting the solution to become essentially Newtonian.

Problem

Instead of approaching the Newtonian solution:

• The fiber barely stretches, the material remains as a thick bulge.
• Even after extending the simulation time, I only obtain something similar to a large droplet/bead, with almost no attenuation.

This behavior remains even when the relaxation time is reduced to 10⁻¹⁰ s, which is surprising because I would expect the viscoelastic model to recover the Newtonian solution in that limit.

 

Has anyone experienced a similar issue with free-surface fiber spinning simulations in Polyflow?

Is there a specific setting, stabilization method, scaling issue, or material parameter that must be adjusted when using differential viscoelastic models?

More specifically, is there any reason why a viscoelastic simulation with a relaxation time of 10⁻¹⁰ s would still produce results that are drastically different from the corresponding Newtonian case?

Any suggestions would be greatly appreciated.

Thanks!
Romain