How to Properly Implement Viscosity in SPH Using ABAQUS Explicit Analysis?

[stevekoja]

Hello everyone,

I have a question regarding the use of viscosity and artificial viscosity in Smoothed Particle Hydrodynamics (SPH) for ABAQUS Explicit dynamic analysis.

Here's the setup I'm working with:

I have a simple wind turbine modeled as a 2-point spring-dashpot system connected to a cylindrical tank. The tank contains fluid, modeled initially as solid elements and then converted into particles using Kernel Interpolation with a PPD value of 2.

The goal of my analysis is to evaluate the system's dynamic response over time with different fluid viscosities.

Parameters I've Tried​

For water:

• Density: 1000 kg/m³
• Dynamic Viscosity: 0.001 Pa·s
• Equation of State (EoS): (c = 1500 m/s, gamma and s = 0)
• Scaling factors for bulk modulus:
  • LinearBulkViscosity = 0.01
  • QuadraticBulkViscosity = 0.1

My input file is fully generated via Python scripting for flexibility.

The Challenge​

Now, I want to simulate oil instead of water, while keeping all other parameters the same except for dynamic viscosity, which I increased to 0.1 Pa·s (100× more viscous than water).

However, I encounter an issue:

1. When I kept the same scaling factors (LinearBulkViscosity = 0.01, QuadraticBulkViscosity = 0.1), the response curve didn’t change at all – unrealistic because oil’s higher viscosity should dampen the system more.
2. When I increased the scaling factors for both linear and quadratic bulk viscosity to 1.0, the results worsened, and the response curve became less realistic overall.

My Question​

How can I correctly implement the role of oil viscosity (0.1 Pa·s) in this SPH analysis? Specifically:

• Should I adjust the artificial viscosity parameters (Linear and Quadratic Bulk)?
• Is there a specific approach to incorporate the dynamic viscosity directly into the SPH simulation in ABAQUS?

I’d really appreciate any suggestions or insights, especially if someone has experience with SPH and viscosity tuning in ABAQUS Explicit.

Thank you!

 

[FEA way]

The documentation chapter about SPH only mentions that the default linear bulk viscosity coefficient (0.06) might need to be lowered when modeling low viscosity fluids such as water because otherwise viscous forces can be too large and result in unrealistic behavior of the fluid.

Another aspect is the choice of the EoS model. Check the documentation chapter Equation of State, paragraph "Modeling Fluids".