Absorbing Boundary Conditions For Seismic Analysis In Abaqus

Introduction

In seismic analysis, accurately capturing the interaction between a structure and the surrounding infinite soil medium is essential for reliable predictions of earthquake response. Absorbing boundary conditions (ABCs) provide a practical way to truncate the computational domain while minimizing artificial reflections of seismic waves back into the model. When using Abaqus for dynamic finite‑element simulations, implementing appropriate ABCs can dramatically improve result fidelity without incurring the prohibitive cost of modeling an entire geological field. This article explains the theory behind absorbing boundaries, outlines the most common ABC formulations available in Abaqus, and presents a step‑by‑step workflow for setting them up in a typical seismic analysis.

Why Absorbing Boundaries Are Needed

The problem of wave reflection

When a seismic wave reaches the edge of a finite model, the numerical algorithm treats the edge as a rigid termination unless special measures are taken. This creates spurious reflections that interfere with the incident wavefield, leading to:

• Over‑estimation of displacement and stress amplitudes.
• Distorted frequency content, especially for low‑frequency components.
• Incorrect assessment of energy dissipation mechanisms such as soil‑structure interaction.

Infinite domain approximation

The Earth can be considered an infinite medium for most engineering‑scale problems. To mimic this condition, absorbing boundaries act as non‑reflective interfaces that allow outgoing waves to leave the model as if they were propagating into an unbounded space. In Abaqus, this is achieved through special boundary definitions that impose a relationship between stress and particle velocity, effectively “absorbing” the wave energy Worth keeping that in mind..

Types of Absorbing Boundary Conditions in Abaqus

Abaqus offers three primary approaches to simulate an infinite domain:

1. Viscous (Lysmer‑Kuhlemeyer) boundaries – the most widely used ABC for 2‑D and 3‑D seismic analyses.
2. Infinite elements (CINFINITE) – specialized elements that extend the mesh into infinity, suitable for high‑frequency wave propagation.
3. Perfectly Matched Layers (PML) – a newer, highly effective technique that damps waves exponentially within a layer surrounding the model.

Each method has its own advantages, limitations, and implementation nuances.

1. Viscous (Lysmer‑Kuhlemeyer) Boundaries

Developed in the 1960s, the Lysmer‑Kuhlemeyer model treats the boundary as a dashpot attached to the soil surface. The dashpot coefficient (c) is defined as:

[ c = \rho , v_s \quad \text{(shear wave)} \qquad c = \rho , v_p \quad \text{(compressional wave)} ]

where (\rho) is the material density, (v_s) the shear‑wave velocity, and (v_p) the compressional‑wave velocity. In Abaqus, this condition is applied through the *DAMPING keyword with the VISCOUS option on the model’s external faces.

Pros

• Simple to set up; requires only material properties.
• Works well for low‑frequency, long‑duration ground motions.

Cons

• Only absorbs normal incidence waves efficiently; oblique incidence leads to partial reflections.
• Not suitable for high‑frequency content (>10 Hz) where numerical dispersion becomes significant.

2. Infinite Elements (CINFINITE)

Abaqus includes the CINFINITE element type (e.g.On the flip side, , CINFINITE6 for 2‑D, CINFINITE8 for 3‑D). These elements are formulated to satisfy the radiation condition analytically, allowing wave energy to propagate outward without reflection.

Key characteristics

Most guides skip this. Don't.

Pros

• Excellent absorption for both normal and oblique waves.
• Handles higher frequencies better than viscous boundaries.

Cons

• More complex mesh generation; the interface must be carefully aligned.
• Slightly higher computational cost due to additional degrees of freedom.

3. Perfectly Matched Layers (PML)

PMLs are artificial absorbing layers surrounding the domain. They work by stretching the coordinate system into the complex plane, causing wave amplitudes to decay exponentially. Abaqus implements PMLs through the *BOUNDARY option PML.

Implementation notes

• Define a PML region with a thickness of at least one to two wavelengths of the highest frequency of interest.
• Choose an attenuation profile (linear, quadratic, or exponential) that matches the expected wave spectrum.

Pros

• Near‑perfect absorption for a wide range of incident angles and frequencies.
• Minimal reflection even for highly transient, high‑frequency content.

Cons

• Requires additional elements, increasing model size.
• Parameter selection (thickness, attenuation function) can be non‑trivial for beginners.

Step‑by‑Step Workflow for Applying Absorbing Boundaries in Abaqus

Below is a practical guide for a 3‑D seismic analysis of a shallow foundation on layered soil, using the viscous Lysmer‑Kuhlemeyer approach. The same logic can be adapted for infinite elements or PMLs Most people skip this — try not to..

Step 1 – Define Material Properties

*Material, name=Soil
*Density
1800.,
*Elastic
1.0e7, 0.3
*Velocity
Vs=250., Vp=500.

• confirm that (\rho), (v_s), and (v_p) are correctly entered, as they directly affect the dashpot coefficients.

Step 2 – Create the Geometric Model

• Build a rectangular block representing the soil domain.
• Extend the block at least 3–5 times the expected wavelength in the direction of wave propagation to reduce boundary influence before the absorbing layer.

Step 3 – Mesh the Model

• Use C3D8R (8‑node reduced integration) elements for the interior.
• Keep the element size ≤ λ/10 for the highest frequency of interest (λ = (v_s/f_{max})).

Step 4 – Apply Viscous Absorbing Boundaries

Identify the faces that will serve as absorbing boundaries (typically the lateral and bottom faces). Then add the DAMPING definition:

*Damping, type=VISCOUS
, 1.0

Replace <face set> with the appropriate set name (e.g., BOTTOM, SIDE1, SIDE2, SIDE3, SIDE4). The 1.0 multiplier is the dashpot coefficient normalized by (\rho v); Abaqus automatically multiplies by the material density and wave speed defined in the material section.

Step 5 – Define the Seismic Excitation

For a ground‑motion input, use a *AMPLITUDE definition and apply it as a *BOUNDARY condition on the base of the model:

*Amplitude, name=GM, definition=TABULAR
0., 0.
0.02, 0.1
...
*Boundary
BOTTOM, 1, 3, GM

• The first column is time, the second is acceleration (or velocity/displacement depending on the analysis type).

Step 6 – Set Up the Dynamic Analysis

*Step, name=Dynamic, nlgeom=YES, inc=1e-4
*Dynamic, Explicit
0., 10., 1e-5
*Output, field, variable=PRESELECT
*Output, history, variable=PRESELECT

• Use an explicit dynamic step for high‑frequency content; otherwise, an implicit dynamic step with appropriate time incrementation works as well.

Step 7 – Run the Simulation and Verify

• After the job completes, examine reaction forces and displacements at points far from the source.
• Plot the time histories of the absorbing faces; a well‑functioning ABC will show minimal reflected wave amplitudes.
• If reflections are noticeable, consider increasing the domain size, refining the mesh, or switching to infinite elements/PMLs.

When to Prefer Infinite Elements or PMLs

Practical Tips for Infinite Elements

• Interface compatibility: The mesh on the solid–infinite element interface must be conformal; use *Tie constraints if necessary.
• Element selection: CINFINITE6 (2‑D) and CINFINITE8 (3‑D) are the standard choices.
• Damping: Even with infinite elements, adding a small amount of numerical damping can improve stability.

Practical Tips for PMLs

• Thickness selection: Choose a thickness of at least 1.5–2 wavelengths of the highest frequency component.
• Attenuation profile: A quadratic profile (α = α_max * (x/L)^2) often provides a good balance between absorption and numerical stability.
• Stability check: Excessive attenuation can cause spurious reflections at the PML–solid interface; perform a sensitivity study.

Frequently Asked Questions

Q1: Can I combine different absorbing boundary types in a single model?

A: Yes. It is common to use viscous boundaries on the lateral faces and infinite elements or PMLs on the bottom to capture deep‑soil effects while keeping the lateral domain compact.

Q2: How do I determine the optimal mesh size for absorbing boundaries?

A: Follow the 10‑elements‑per‑wavelength rule for the highest frequency of interest. For a shear‑wave velocity (v_s) and maximum frequency (f_{max}), the wavelength is (\lambda = v_s / f_{max}). Mesh size (h ≤ λ/10) Simple, but easy to overlook..

Q3: Will absorbing boundaries affect the static response of the model?

A: No. ABCs are formulated to act only on dynamic terms (stress–velocity relation). In a purely static analysis, they have no effect.

Q4: Is it necessary to calibrate absorbing boundary parameters against analytical solutions?

A: For critical projects, validating the ABC implementation against a known analytical or semi‑analytical solution (e.g., a half‑space Green’s function) is recommended. This ensures that reflections are within acceptable limits And that's really what it comes down to..

Q5: What are the computational cost differences among the three ABC types?

A:

• Viscous boundaries: Minimal extra cost (only dashpot forces).
• Infinite elements: Approximately 10–20 % increase in DOFs compared to a model without them.
• PMLs: Can double the element count if a thick layer is used, but the improved accuracy may justify the expense.

Conclusion

Absorbing boundary conditions are indispensable tools for realistic seismic analysis in Abaqus. By preventing artificial wave reflections, they enable engineers to simulate the infinite nature of the earth with a finite computational domain. The viscous Lysmer‑Kuhlemeyer approach offers simplicity and speed for low‑frequency problems, while infinite elements and Perfectly Matched Layers provide higher fidelity for complex, high‑frequency, or obliquely incident wavefields And it works..

Implementing ABCs follows a clear workflow: define accurate material properties, size and mesh the domain appropriately, select the most suitable boundary type, and verify the absence of reflected energy. With these steps, practitioners can achieve reliable, trustworthy seismic predictions that support safe, cost‑effective design of structures ranging from bridges and tall buildings to offshore platforms.

Incorporating absorbing boundaries not only aligns with best practices in computational seismology but also enhances the credibility of simulation results, making them a cornerstone of modern earthquake engineering analyses in Abaqus.