1 Fundamentals of Free Vibration
1.1 Definition and basic assumptions
Free vibration is the motion of a mechanical or structural system after it has been disturbed (for example, by an initial displacement, an initial velocity, or a short transient input) and then allowed to evolve without continued external forcing. The subsequent time history is determined by the system’s intrinsic properties and constraints, typically including mass distribution, stiffness, and energy dissipation mechanisms. In standard analyses, assumptions often include linear behavior, small displacements, and constant system parameters during the observation interval.
1.2 Idealized models (single degree of freedom)
Many introductory treatments use a single degree of freedom (SDOF) idealization, where the system is represented by an equivalent mass, spring-like stiffness, and damping element connected in parallel or series. The physical motion is described by one generalized coordinate (e.g., displacement of a point). This reduction clarifies how natural frequency and damping shape the oscillation, while multi-degree-of-freedom reality is addressed later through modal methods.
1.3 Harmonic motion and system restoring forces
In linear SDOF systems, restoring forces are modeled as proportional to displacement (spring behavior). For an undamped system, the restoring force leads to oscillation at the natural frequency determined by the ratio of stiffness to inertia. When damping is present, the same restoring action persists but the oscillation loses energy over time, producing a decaying amplitude while retaining characteristic timing (period) related to the system parameters.
1.4 Role of damping and energy dissipation
Damping represents mechanisms that convert mechanical energy into heat or other forms of dissipation. It is central to interpreting measured free-decay responses because it governs how rapidly oscillation amplitude decreases. Depending on damping strength, the time response may show smooth return to equilibrium without overshoot, or it may oscillate while gradually shrinking. In practice, measured damping includes contributions from material hysteresis, friction, aerodynamic effects, and boundary losses.
2 Governing Equations
2.1 Undamped vibration
An undamped SDOF system is governed by the second-order differential equation \[ m\ddot{x}+kx=0, \] where \(m\) is mass, \(k\) is stiffness, and \(x\) is displacement. Solutions are harmonic and can be expressed in sinusoidal or exponential form, with constant amplitude because no energy is dissipated.
2.1.1 Natural frequency relationships
For the undamped SDOF model, the natural (angular) frequency is \[ \omega_n=\sqrt{\frac{k}{m}}, \] and the cyclic frequency is \(f_n=\omega_n/(2\pi)\). In measurement contexts, these relationships motivate how changing stiffness or added mass shifts observed resonant or free-decay frequencies. For complex structures, analogous natural frequencies arise from eigenvalue problems in multi-degree-of-freedom formulations.
2.2 Damped vibration
With viscous damping, a standard SDOF model becomes \[ m\ddot{x}+c\dot{x}+kx=0, \] where \(c\) is the damping coefficient. The response is typically a decaying oscillation or a non-oscillatory decay, depending on the damping magnitude.
2.2.1 Underdamped, critically damped, overdamped cases
- Underdamped: The system oscillates while amplitude decays exponentially. This is common for many real structures at low-to-moderate damping levels.
- Critically damped: The system returns to equilibrium as quickly as possible without oscillating.
- Overdamped: The return to equilibrium is slower and monotonic.
These cases correspond to whether the damping ratio \(\zeta\) is less than, equal to, or greater than one, where a typical definition is \(\zeta=c/(2\sqrt{mk})\).
2.3 Forced vs. free vibration distinction
In forced vibration, external inputs continually add energy (harmonic forcing, shocks, or sustained excitation), so the system’s response is shaped by the driving spectrum as well as by the system dynamics. Free vibration, in contrast, is analyzed after forcing ceases, and its evolution reflects the system’s homogeneous dynamics. This distinction matters experimentally because data windows must isolate portions dominated by free decay rather than ongoing excitation or re-excitation.
2.4 Initial conditions and response forms
Free vibration solutions depend on initial displacement \(x(0)\) and initial velocity \(\dot{x}(0)\). For the underdamped case, the displacement can be written as a decaying envelope multiplied by a sinusoid with a damped natural frequency. The measured waveform phase and amplitude at the start of observation therefore encode the initial conditions, even when system parameters are identical.
3 Natural Frequencies and Mode Shapes
3.1 Single-mode behavior
If a system is excited in a way that predominantly activates one dynamic pattern, the free response can appear nearly sinusoidal with a single decay rate and period. In SDOF systems, this is exact; in real structures, it is an approximation valid when other modes contribute minimally within the measured time span. The apparent frequency observed during early decay is often most informative when modal contributions are separable.
3.2 Multi-degree-of-freedom systems
Real structures require multiple coordinates and are modeled by coupled equations of motion. Natural frequencies emerge as eigenvalues of the system’s mass and stiffness matrices (and damping, depending on model detail). The corresponding mode shapes describe how different parts of the structure move relative to one another when vibrating at a particular natural frequency.
3.3 Modal decomposition concepts
Modal decomposition expresses the system response as a sum of modal coordinates, each associated with a natural frequency and a damping behavior. Under linear assumptions, each mode’s coordinate obeys a decoupled second-order equation in the modal space. In measurement practice, extracted parameters often correspond to dominant modes, with the remaining modes treated as background contributions or residual effects.
3.4 Frequency response interpretation (conceptual)
While free vibration is a time-domain phenomenon, its modal content can be interpreted in the frequency domain. Peaks in spectral representations correspond to resonant components associated with natural frequencies. In decays, the spectrum reflects both the modal frequencies and the damping-induced spread (broader peaks for higher damping). Conceptual frequency response interpretations help guide peak identification, especially when multiple modes are present.
4 Time-Domain Analysis
4.1 Free-decay signals
A free-decay signal is the measured time series after external excitation has ceased. It may be captured by displacement, velocity, or acceleration sensors. Proper identification of the decay segment is essential because early portions may include transient effects, and late portions may be dominated by noise. The core goal is to relate the observed waveform to exponential (or exponential-modulated sinusoidal) forms predicted by the governing equations.
4.2 Estimating frequency from waveform characteristics
A common approach to frequency estimation uses the period between successive peaks (or zero crossings) in the decay. For underdamped motion, the oscillation frequency remains relatively stable even as amplitude decays, allowing the extraction of a damped frequency. Converting to an undamped natural frequency may be possible when damping is estimated, using relationships between damped and undamped frequencies.
4.3 Logarithmic decrement for damping
Logarithmic decrement quantifies damping by measuring how much the oscillation amplitude decreases over a fixed number of cycles. For an underdamped response, the envelope decays exponentially, making the logarithmic decrement directly related to the damping ratio. This method is robust when peaks are clearly distinguishable and when damping remains approximately constant over the observation window.
4.3.1 Practical measurement procedure overview
A typical workflow is:
- Identify a sequence of peaks in the free-decay portion.
- Measure amplitudes of two peaks separated by a known integer number of cycles.
- Compute the logarithmic decrement as the natural log ratio of amplitudes.
- Convert the decrement to damping ratio using the standard SDOF relations.
- Repeat with multiple peak pairs to assess consistency.
4.4 Decay envelope and amplitude tracking
Rather than relying solely on discrete peaks, amplitude tracking estimates the decay envelope across time. Methods may fit an exponential function to rectified or Hilbert-transformed signals. Envelope fitting can improve robustness when noise obscures peak locations. It also supports simultaneous estimation of frequency (via oscillatory component) and damping (via exponential envelope), though model assumptions must remain appropriate.
5 Frequency-Domain Analysis
5.1 Transforming measured signals (conceptual)
Frequency-domain analysis typically starts with a transform (such as a Fourier transform) to convert time series into a spectral representation. Conceptually, a free-decay signal contains a concentration of energy around modal frequencies, with the spread determined by damping and observation duration. While the transform itself is mathematical, practical results depend strongly on preprocessing, windowing, and the chosen analysis segment.
5.2 Peak picking for resonant components
Spectra of decay signals often show one or more dominant peaks corresponding to natural frequencies. Peak picking uses local maxima to estimate frequencies and sometimes amplitudes. When multiple modes are close, peaks may overlap, requiring more sophisticated fitting or time-frequency methods. For accurate parameter extraction, the analyst must ensure the frequency resolution is sufficient to separate modal contributions.
5.3 Bandwidth, resolution, and leakage (conceptual)
- Resolution relates to how narrowly the analysis can distinguish nearby frequencies; it improves with longer observation windows but can degrade when the signal amplitude becomes too small relative to noise.
- Bandwidth is influenced by damping and the finite length of data; higher damping tends to broaden spectral features.
- Leakage occurs when the analysis window does not align with integer periods of the underlying oscillations, causing energy spread across bins. Windowing reduces leakage but changes the effective spectral shape, influencing peak height and width measurements.
5.4 Identifying dominant modes
Dominant modes are those contributing most strongly to the measured response under the given excitation and sensor placement. Identification may rely on peak prominence in spectra, consistency across multiple tests, or coherence across channels when multiple sensors are used. In multi-mode situations, interpreting free-decay spectra requires caution because modal coupling and sensor sensitivity can emphasize specific components over others.
6 Parameter Identification for Measurement
6.1 Determining natural frequency
Natural frequency estimation uses time-domain or frequency-domain evidence. In time-domain methods, peak spacing provides damped frequency estimates; these may be translated to undamped natural frequency when damping is known. In frequency-domain methods, spectral peak locations provide frequency estimates that must consider resolution and leakage effects. Parameter identification is best supported by selecting decay intervals where model assumptions approximate reality.
6.2 Estimating damping ratio
Damping ratio estimation can be performed with logarithmic decrement, envelope fitting, or curve-fitting of the full decaying response. The selected method should reflect signal quality: peak-based decrement works well for clean oscillations, while envelope or global fitting can handle noise and amplitude variations. For higher damping, the response may deviate from simple sinusoidal decay, requiring careful model selection.
6.3 Using system models to fit data
Parameter identification often uses a model-based fit to measured displacement, velocity, or acceleration. A fit may include unknown frequency, damping ratio, initial phase, and amplitude scale. For multi-degree-of-freedom systems, modal fitting may use superposition models and compare predicted time histories with measured data. The model’s validity depends on linearity, constant parameters, and correct selection of the free-decay segment.
6.4 Uncertainty and confidence in extracted parameters
Measured parameters are subject to uncertainty from noise, window selection, sensor calibration, and model mismatch. Confidence evaluation can use repeated trials, bootstrap resampling, or sensitivity analysis with respect to analysis settings. Uncertainty reporting is important because small differences between tests might reflect processing choices rather than changes in structural dynamics.
7 Experimental Setup and Instrumentation
7.1 Excitation methods that start free vibration
Free vibration tests require a disturbance that initiates oscillation but then stops. Common approaches include impact excitation, release mechanisms, and using an actuator for a short pulse followed by complete removal of force. The excitation method should be repeatable and should produce a usable decay segment without ongoing contact, re-excitation, or significant nonlinear behavior at the start.
7.2 Sensors: accelerometers, displacement transducers, velocimeters
- Accelerometers are widely used because they provide direct acceleration signals with broad frequency coverage. They require proper integration strategy if displacement or velocity is needed.
- Displacement transducers offer direct position measurement but may have limited bandwidth or require careful mounting and alignment.
- Velocimeters measure velocity directly and can reduce integration issues, though instrumentation availability and operating ranges vary.
Sensor placement influences which modes are observed and can affect apparent damping and frequency estimates.
7.3 Sampling considerations (conceptual)
Sampling frequency must be high enough to capture the highest expected modal frequency content without aliasing. Additionally, the sampling rate should be consistent across repeated tests to enable comparability. Time alignment and data acquisition timing accuracy can matter when extracting phase-sensitive quantities or when multiple channels are analyzed for modal decomposition.
7.4 Noise, filtering, and signal conditioning (conceptual)
Noise can obscure peaks and distort envelope estimation. Signal conditioning may include amplification, anti-alias filtering, and careful gain selection to avoid saturation. Filtering must be used judiciously: overly aggressive filtering can alter decay shape and bias damping estimates. Baseline correction, detrending, and consistent preprocessing steps help ensure that extracted parameters reflect system dynamics rather than measurement artifacts.
8 Data Processing Workflow
8.1 Preprocessing and detrending
Before analysis, signals may require detrending to remove offsets and slow drift. Preprocessing can also correct for sensor scaling factors and ensure correct units. If integration is used to obtain displacement from acceleration, drift control becomes a key step, often requiring high-pass filtering or other strategies consistent with the analysis objectives.
8.2 Windowing and segment selection
Selecting the correct time segment for “free decay” is critical. Analysts typically choose a start time after excitation has ceased and an end time before noise dominates. Windowing in the frequency domain reduces leakage but should be selected to balance spectral fidelity against amplitude bias. Segment selection choices can significantly influence peak widths, envelope fits, and thus inferred damping.
8.3 Peak detection and fitting strategies
Peak detection involves identifying local maxima or minima under noise. The algorithm choice affects robustness, especially when oscillation amplitude becomes small. Fitting strategies may include:
- fitting a decaying sinusoid to the time history,
- fitting an exponential envelope,
- performing multi-exponential or multi-modal fits for complex responses.
Good fits should produce residuals consistent with noise assumptions and should not systematically over- or under-predict early or late portions.
8.4 Validation checks (residual behavior, repeatability)
Validation helps confirm that the selected model and analysis settings are appropriate. Residuals—differences between measured and fitted responses—should show no clear periodic structure if the model captured the dominant modes. Repeatability across tests provides additional evidence that parameter estimates are stable and that excitation and measurement conditions are consistent.
9 Common Applications of Free-Vibration Measurement
9.1 Modal testing concepts
Modal testing aims to determine dynamic properties such as natural frequencies, mode shapes, and damping characteristics. Free-vibration measurements contribute by providing decay data that reveals modal parameters without ongoing forcing. Depending on the test approach, free-decay information may be used alone or combined with other excitation types to improve modal coverage.
9.2 Condition monitoring and health assessment (general)
Changes in natural frequencies and damping can indicate alterations in stiffness, mass, connectivity, or frictional losses. In general condition monitoring practice, free-vibration tests serve as a baseline comparison over time. However, interpretation requires careful control of environmental and operational factors that might change apparent dynamics without structural damage.
9.3 Vibration-based quality control (general)
In quality control, manufacturers may use free-vibration tests to verify consistency across manufactured components. Variations in mass distribution, material properties, or assembly stiffness can shift modal frequencies and damping behavior. Because free-decay analysis can be time-efficient, it is often suitable for screening and acceptance testing where rapid measurement is valuable.
9.4 Calibration and model verification
Free-vibration measurements also support calibration of analytical models and parameter tuning. By comparing predicted and measured decay responses, analysts can adjust model parameters (such as stiffness distributions or damping assumptions) to improve predictive accuracy. Calibration helps ensure that subsequent simulations or design decisions rest on validated dynamic behavior.
10 Special Cases and Practical Limitations
10.1 Nonlinearity effects on decay and frequency drift
Real systems may exhibit nonlinear stiffness, frictional damping, or amplitude-dependent behavior. In such cases, the frequency can vary as the oscillation amplitude decays, producing drift in peak spacing or changing spectral peak shapes. Damping may also become amplitude-dependent, causing logarithmic decrement values to vary over the decay interval. Analysts must recognize when a linear decaying-sinusoid model is insufficient.
10.2 Coupling and boundary-condition sensitivity
Coupled modes arise when multiple degrees of freedom interact, especially near mode crossings or when boundary conditions are uncertain. Small differences in support conditions, mounting torque, or fixture stiffness can alter measured frequencies and damping. Mode coupling can also cause non-single-exponential decays, leading to composite decay envelopes or multiple apparent frequencies in the same time window.
10.3 Environmental influences (temperature, loading history—general)
Environmental factors such as temperature can affect material stiffness and damping. Loading history can influence frictional contacts or material behavior, changing how energy is dissipated. In measurement programs, controlling or logging such factors supports distinguishing between true dynamic changes and measurement-to-measurement variability.
10.4 Limitations of simplified SDOF interpretation
An SDOF approach is most reliable when one mode dominates and when damping behavior is approximately linear and constant. In multi-mode systems, fitting a single decaying sinusoid may produce biased estimates—frequency may shift toward an effective value, and damping may appear different because multiple modes contribute to the observed signal. In such cases, multi-modal modeling or more comprehensive modal identification methods are needed.
11 Safety, Accuracy, and Reporting
11.1 Ensuring safe excitation levels (general)
Excitation must be controlled to avoid damage to equipment, sensors, or test articles. Even when force levels are modest, repeated impacts or actuator pulses can create hazards in handling and setup. Safe practice includes secure mounting, protected sensor cabling, and limits on excitation magnitude consistent with structural integrity and measurement objectives.
11.2 Documentation of test conditions
Accurate reporting requires documenting excitation type, locations of disturbance and measurement, support or mounting configuration, and duration of the recorded free-decay segment. Recording metadata also includes sensor make/model, sampling rate, and preprocessing steps. Such documentation is crucial for interpreting results and enabling comparisons across teams or time.
11.3 Reporting conventions for frequency and damping
Results are typically reported as natural frequencies (often in Hz or rad/s) and damping metrics (commonly damping ratio or logarithmic decrement). If frequency is estimated as a damped frequency, conversions to undamped natural frequency should be stated along with the assumptions used. Clear units, sign conventions, and definitions reduce ambiguity.
11.4 Reproducibility and traceability of measurements
Reproducibility depends on repeatable excitation, consistent sensor placement, and standardized data processing. Traceability is supported by calibration records for sensors and acquisition systems, and by recording software versions and parameter settings used in analysis. Together, these practices help ensure that extracted free-vibration parameters are reliable and comparable.