Operational Modal Analysis
by Nikolay Donets
– March 1, 2012 · 4 min read
In this article, we will review the basic concepts of Operational Modal Analysis OMA and we will present the most common OMA techniques, namely: the Frequency Domain Decomposition FDD method, and the Stochastic Subspace Identification SSI method. We will also introduce the basics of the system identification method, which is used in OMA to estimate the modal parameters of a structure. OMA does not require special excitation of a structure and this is the main advantage of this method.
What is Operational Modal Analysis
Operational modal analysis is used in the field of experimental modal analysis to identify the modal parameters of a structure using only response measurements. The structure is assumed to be excited by unknown random forces and the response is measured at several locations on the structure. The unknown excitation is modeled as white noise and the response is used to estimate the modal parameters of the structure.
A structure will vibrate at a certain natural frequency when it is subjected to an excitation at that frequency. Natural frequency is the frequency at which a structure will vibrate when it is not subjected to damping.
Damping is the dissipation of energy that occurs when a structure is subjected to vibration. The amount of damping in a structure can be represented by the damping ratio.
Mode shape is a plot of the amplitude of vibration at each measurement point at a particular natural frequency. A mode shape is a plot of the vibration pattern of a structure at a particular natural frequency.
Operational modal analysis OMA is a method for extracting modal parameters of a structure from output-only measurements, i.e. without measuring the excitation forces. In OMA, the input force to the structure is not measured but rather assumed to be a zero-mean white-noise process. The main assumptions in OMA are the following:
- The input force to the structure is a zero-mean white-noise process: the input force is a random process with constant Power Spectral Density PSD over all frequencies. This assumption is required to ensure that the Cross-Spectral Density CSD matrix of the response is proportional to the Frequency Response Function FRF matrix.
- The input force is spatially uncorrelated with itself: the force at one point on the structure is uncorrelated with the force at any other point. The assumption is important because the spatial correlation of the input force will affect the CSD matrix of the response
- The input force is uncorrelated with the response of the structure: the input force is not correlated with the response of the structure since the correlation between the input force and the response will affect the CSD matrix of the response as well
- The structure is lightly damped: the damping ratio of the structure is small
The last three assumptions are generally satisfied for many structures. For example, bridges usually have low damping ratios. The first assumption is satisfied in the case of ambient vibration, in which the input force to the structure is the wind and the traffic loads. The ambient vibration can be assumed to be a zero-mean white-noise process if the wind speed is constant over the duration of the measurement and if the traffic loads are uniformly distributed over the bridge deck.
The main advantage of OMA is that the modal parameters can be identified without measuring the input forces. OMA is especially useful for structures that are difficult to excite, e.g. offshore structures, or for structures that are excited by ambient forces, e.g. bridges.
The main disadvantage of OMA is that the modal parameters can only be identified for the modes that are excited by the unknown input forces. The modal parameters of the unexcited modes cannot be identified.
Frequency Domain Decomposition Method
The Frequency Domain Decomposition FDD method is a simple OMA method that can be used to identify the natural frequencies, the damping ratios, and the mode shapes of a structure.
The FDD method is based on the on the fact that the Singular Value Decomposition SVD of the output Power Spectral Density PSD matrix can be used to estimate the natural frequencies, the damping ratios, and the mode shapes of a structure. The FDD method involves the following steps:
- Compute the output PSD matrix
- Compute the Singular Value Decomposition SVD of the PSD matrix
- Plot the singular values of the PSD matrix versus the frequency
- The natural frequencies and damping ratios are estimated from the peaks of the singular values
- The mode shapes are estimated from the first singular vector at the natural frequencies
Stochastic Subspace Identification Method
The Stochastic Subspace Identification SSI method is a more advanced OMA method than the FDD method. The SSI method can be used to estimate the natural frequencies, the damping ratios, and the mode shapes of a structure.
The SSI method involves the following steps:
- Construct the block Hankel matrix from the measured data
- Perform the singular value decomposition of the Hankel matrix
- Perform a projection to obtain the Extended Observability matrix
- Perform a projection to obtain the state sequence
- Perform a singular value decomposition to obtain the system matrices and calculate modal parameters
- An additional step is to vary the model order and plot a stabilisation diagram: grouping together identified modal parameters helps to remove noisy estimations