Numerical Integration in Structural Dynamics

Structural dynamics is the study of the response of structures to dynamic loads, such as those caused by wind, waves, machinery, or seismic activity. Unlike static analysis, where loads are constant, dynamic analysis must account for the time-varying nature of forces and the resulting inertia and damping effects.

The core of linear structural dynamics is the Equation of Motion. For a Multi-Degree-of-Freedom (MDOF) system, this equation is typically expressed in matrix form:

Here, M represents the mass matrix, C is the damping matrix, and K is the stiffness matrix. The vectors {u}(t), {u}(t), and {u}(t) represent acceleration, velocity, and displacement, respectively, while F(t) is the external dynamic force vector.

While exact analytical solutions exist for simple Single-Degree-of-Freedom (SDOF) systems, complex MDOF systems with nonlinear material properties or geometry require Numerical Integration Methods. These methods, also known as time-stepping or direct integration methods, approximate the response of the system at discrete time intervals.

Time Stepping Fundamentals

Direct integration methods do not attempt to find a mathematical function that satisfies the dynamic equation for all time. Instead, they satisfy the equation at specific discrete time steps: t, t + t, t + 2t, etc. The goal is to calculate the state of the structure (displacement, velocity, acceleration) at time t + t based on the known state at time t.

These methods generally fall into two categories: Explicit methods and Implicit methods.

Explicit Integration Methods

Explicit methods calculate the state at the end of a time step using only the known values from the beginning of the step. The most prominent example in structural dynamics is the Central Difference Method.

The Central Difference Method

This method approximates velocity and acceleration using finite difference equations derived from Taylor series expansions. The acceleration at time t is approximated by:

By substituting these approximations into the equations of motion, the displacement at t+t can be solved directly without inverting the stiffness matrix, provided the mass matrix is diagonal (lumped mass).

Pros and Cons:

• Advantage: Computationally efficient per step because no matrix inversion is required.
• Disadvantage: It is conditionally stable. To prevent the solution from diverging (growing infinitely), the time step t must be smaller than a critical value related to the system's natural period (T_n). Specifically, t must be less than T_n/.

This stability constraint makes explicit methods suitable for wave propagation problems (like high-impact explosion analysis) where the duration of interest is very short, or for nonlinear problems where the stiffness changes rapidly.

Implicit Integration Methods

Implicit methods use the equilibrium conditions at time t + t to solve for the state at t + t. This implies that the stiffness matrix enters the calculation, and a system of linear equations must be solved at every time step.

The Newmark-Beta Method

Proposed by Nathan M. Newmark in 1959, this is arguably the most widely used method in structural dynamics software. It introduces two parameters, and , which control the accuracy and numerical stability of the integration.

The method uses the following assumptions for velocity and displacement at t + t:

A common choice is the Average Acceleration Method, where = 1/2 and = 1/4. This specific combination renders the method unconditionally stable.

Pros and Cons:

• Advantage: Unconditional stability allows for much larger time steps than explicit methods. The time step is chosen based on accuracy requirements rather than stability limits. This is ideal for low-frequency structural response, such as buildings swaying during an earthquake.
• Disadvantage: Computationally more expensive per step because the inverse of the effective stiffness matrix must be calculated (or a linear system solved). In nonlinear analysis, this matrix changes and must be updated/inverted at every step.

Hilber-Hughes-Taylor (HHT) Method

The HHT-alpha method is an extension of the Newmark-Beta method designed to introduce a small amount of numerical damping. The Average Acceleration method (standard Newmark) preserves energy indefinitely, which can be problematic in long-duration analyses where spurious high-frequency noise might accumulate. HHT damps out these artificial high-frequency vibrations while retaining accurate response in the lower frequency range.

Stability, Accuracy, and Convergence

When selecting a numerical integration method for structural dynamics, one must analyze three key properties:

1. Stability: Will the numerical solution remain bounded for any time step, or is there a limit (t_crit)? Implicit methods generally win here for structural problems.
2. Accuracy: How close is the numerical approximation to the true mathematical solution? Accuracy is usually dictated by the size of t. Smaller steps capture higher frequencies more accurately but increase computation time.
3. Convergence: As t approaches zero, does the numerical solution approach the exact solution? All standard methods used in commercial software (like ANSYS, Abaqus, or SAP2000) are convergent.

Conclusion