1. Listening to the Machine
Every structure has a voice, the set of frequencies at which it rings when it is struck, and the engineer who understands that voice can fix a vibrating machine, isolate a sensitive instrument or defend a structure against an earthquake before the damage is done. Vibration testing is the systematic way of making a structure speak: the engineer excites the structure with a known force, measures the response, and draws the map of its resonances, the natural frequencies, the mode shapes and the damping that govern how it moves. The companion discipline, modal analysis, turns those measurements into a mathematical model, a set of modes that describes the structure’s vibration comprehensively enough to predict its behavior under any loading.
This article develops the full path from the vibrating machine to the validated model: the equations of the single degree of freedom and the multi degree of freedom system, the measurement of the frequency response, the excitation methods from the hammer to the shaker, the estimation and the presentation of the modes, and the many engineering uses of the validated model, from the troubleshooting of a resonant machine to the design validation of a lightweight, dynamic structure. The article closes with a worked interpretation, showing how the same data that diagnoses the problem also feeds the fix.
Vibration is not a mysterious failure but a measured property: the frequency, the amplitude and the phase of a structure’s motion are as measurable as its temperature, and modal analysis is the language in which those measurements are translated into design change.
2. The Single Degree of Freedom and the Resonance
The simplest vibrating system, and the seed of all vibration theory, is the single degree of freedom system: a mass on a spring with a damper, the mass able to move in only one direction. The system has a natural frequency, the rate at which it swings freely when displaced and released, set by the stiffness of the spring and the size of the mass, and a damping ratio, the fraction of the critical damping that the damper provides, which governs how quickly the free swings die away and how large the forced response grows. When the structure is driven by a periodic force whose frequency equals the natural frequency, the response grows, in the undamped case without limit, in the damped case to a peak whose height is controlled by the damping ratio; and the single degree of freedom system tuned close to resonance is the classic vibration failure, the machine that shakes itself to pieces at its own natural frequency.
The amplification at resonance is quantified by the quality of the resonance: a lightly damped system, one whose damping ratio is small, has a high, sharp resonance peak and a tremendous amplification, while a heavier damped system spreads and lowers the peak. The engineer controls a vibration problem at resonance by separating the forcing frequency from the natural frequency, by raising the stiffness or by adding mass to move the resonance away, by adding damping to cap the amplification, or by isolating the source, and the single degree of freedom model, simple as it is, is the whole toolbox for the most common vibration trouble, the coincidence of the excitation with a resonance.
The measured data of vibration testing, the frequency response function, is exactly the picture of the resonance: the peak of the response at each natural frequency, the width of the peak giving the damping, the phase of the response, falling through ninety degrees as the frequency crosses the resonance, marking the point precisely, and the engineer reads the natural frequency and the damping straight off the plot. The single degree of freedom is the alphabet of vibration, and the whole machinery of modal analysis is the combination of this alphabet, one mode at a time, into the full vocabulary of a real, many mode structure.
3. From One Mode to Many
A real structure, a machine tool column, a gearbox housing, a printed circuit board, is not a single mass and spring but an infinite assembly of degrees of freedom, and its motion is the superposition of a large number of modes, the mode shapes and natural frequencies at which the whole structure tends to vibrate. Each mode is a pattern: the structure bends or twists in a characteristic shape, with some points at rest, the nodes, and others moving the most, the antinodes, and the mode is defined by its shape, its frequency and its damping, the same three numbers that describe the single degree of freedom system, one triple for every mode across the frequency range of interest. The first few modes, the lowest frequency bending and torsion shapes, are the ones that dominate the response and the ones that cause most of the trouble, while the higher modes, denser and harder to excite, matter for the broadband high frequency vibration and the acoustic noise.
The engineering value of the modes is in the superposition: any vibration of the structure, whatever the forcing, is the sum of the mode shapes, each mode participating to a degree set by how well the forcing matches its shape and its frequency. The mode shape is therefore the key to the fix: if the measured vibration is dominated by the first bending mode, the engineer knows where the structure bends most, the antinode, and knows which change, a rib there, a mass there, a damper there, moves or damps that mode; and the modal model, the complete set of modes, allows the response to any excitation to be predicted without further measurement. The many mode structure is comprehensible only through its modes: the thousands of measured response points reduce to a handful of modal triples, and the engineer thinks in the compressed language of the mode shapes, bending, torsion, breathing, and treats the vibration by acting on the few modes that rule.
The experimental modal analysis is the measurement of these modes: the excitation of the structure over the frequency range, the measurement of the response at a grid of points, and the fitting of the modal parameters, the frequency, the damping and the shape of each mode, to the measured frequency response functions, giving the validated modal model of the real structure, the fingerprint the finite element model is tuned to match.
4. Exciting and Measuring the Structure
Vibration testing needs two instruments: a way to put a known force into the structure and a way to measure the resulting motion. The excitation comes in two standard forms. The impact hammer, a hammer with a force transducer in its tip, strikes the structure a sharp blow and delivers a broad band of frequencies in a single tap, the frequency range set by the hardness of the tip and the mass of the hammer, and is ideal for the quick, portable test of a small or medium structure. The electrodynamic shaker, an electromagnetic actuator driving the structure through a stinger and a force cell, delivers a controlled, sinusoidal or random force over a long, repeatable test, and is the tool for the larger structure, the precise measurement and the multi input test where the direction of the excitation must be fully controlled.
The response is measured by the transducers: the accelerometer, a small mass on a piezoelectric crystal, now the standard sensor, measures the acceleration of the structure at its mounting point, its small size and its broad frequency range letting the engineer attach a grid of sensors and measure the mode shapes; the laser vibrometer, counting the Doppler shift of a reflected laser, measures the velocity without touching the structure, ideal for the light, delicate or hot component that a glued accelerometer would alter. The measured signals pass through the conditioning and the analyzer, which sample the time histories, convert them to the frequency domain with the fast Fourier transform, and assemble the frequency response functions: the ratio of the response to the force at every frequency, from which the mode is extracted.
The test quality is set by the measurement discipline: the calibration of the force cell and the accelerometers, the analysis of the coherence, the measure of how much of the response is caused by the applied force rather than by the noise, the averaging of the repeated measurements, and the choice of the frequency resolution and the window, because the frequency response function is only as trustworthy as the sampling that built it. The counting of the structural response across its frequency range, hammer and shaker, accelerometer and laser, is the raw listening from which the modes are estimated.
5. Estimating the Modes from the Data
The measured frequency response functions contain the modes buried in their peaks and their phase transitions, and the modal parameter estimation pulls them out. The estimator fits a mathematical model, usually the sum of the single degree of freedom resonance terms, each mode’s frequency, damping and residue, to the measured functions, and the curve fit, manual or automated, finds the triple for every visible mode across the frequency range. The mode shape is assembled from the residues: the response measured at each grid point in the same mode gives that point’s motion in that mode, and the collection of the points, normalized and arranged, becomes the animation of the structure bending, twisting or breathing at its natural frequency, the mode shape that the engineer watches rotate on the screen.
The estimation is a craft as much as an algorithm. The closely spaced modes, two resonances nearly overlapping, are the hardest to separate, and need the higher resolution measurement and the more sophisticated multi degree of freedom fitter. The weakly excited modes, modes whose shape the applied force barely stirred, show only a small peak and are easily lost in the noise, and the engineer often repeats the test with the excitation moved to the point where the mode is strongly excited. The damping estimate, read from the width of the resonance peak, is the least accurate of the three modal numbers, and the modal model’s damping values are quoted with the caveat that they depend on the load, the temperature and the assembly state of the structure.
The validation of the estimated model closes the loop: the synthesized frequency response, recomputed from the fitted modes, is overlaid on the measured one, and the agreement of the peaks, the frequencies and the phase across the range confirms that the modes truly captured the structure. The cross validation, the measurement of a response point not used in the fit compared with the model’s prediction, and the modal assurance criterion, the comparison of two estimates of the same mode shape, check the consistency of the model, and the validated modal model, the complete collection of the frequencies, the damping values and the animated shapes, is handed to the engineer as the structure’s true dynamic description.
6. Using the Modal Model
The validated modal model is the working tool of modern vibration engineering. In the design office, the experimental modes are the benchmark for the finite element model: the analyst runs the modal analysis of the finite element model, compares the computed frequencies and mode shapes with the measured ones, and tunes the model parameters, the stiffness and the boundary conditions, until the analytical and the experimental modes agree, the model correlation and the updating that make the simulation trustworthy enough to predict the stresses and the fatigue under loads that cannot be tested. The numerical model that matches the measured modes is the model whose predicted stresses and resonances the engineer can trust, and the correlation of the modes is the currency that buys that trust.
In the plant, the modal model is the diagnosis of the vibrating machine. The measured operating vibration, the response of the machine under its working load, is decomposed into the modes, and the engineer sees which mode is dominant and where its antinode lies, and knows then the fix: separate the resonance, stiffen the antinode, add the mass or the damper in the position where the mode moves most, or isolate the excitation at its source, and the same measured data that found the problem predicts the effect of each proposed change without trial and error. The mode shape also guides the placement of the sensors and the vibration isolators, the sensors where the mode is largest for the clearest signal, the isolators at the nodes where the motion is least, and the avoidance of a resonance in a rotating structure, the critical speed of a rotor, is engineered by comparing the rotor’s excitation spectrum with the structure’s measured modal frequencies and moving one away from the other.
The modal model also serves the acceptance and the monitoring: the measured resonance frequencies of a product line become the acceptance fingerprint that every new unit is compared against, catching the missing bolt or the void in the casting that changes a mode, and the ambient and the operating vibration of a structure, monitored over time, are decomposed into the modes to track the wear, the loosening or the cracking that shifts a natural frequency, the structural health monitoring that reads the health of a machine from the drift of its modes. The modal model is not the end of the test but the beginning of its uses, the bridge between the measured vibration and the engineered fix.
7. Reading the Test as a Fix
7.1 The standard test workflow
- Define the frequency range and the grid of measurement points
- Calibrate the force cell and the accelerometers, mount the grid
- Excite with the hammer or the shaker, measure the response
- Build the frequency response functions with the good coherence
- Estimate the frequencies, the damping and the mode shapes
- Validate the model against the independent response measurements
- Use the validated modes for the fix, the model update or the monitoring
7.2 Excitation and measurement vocabulary
| Instrument | Role in the test |
|---|---|
| Impact hammer | broad band force in a single tap |
| Electrodynamic shaker | controlled, repeatable force over the range |
| Force cell | measures the applied force |
| Accelerometer | measures the response at the grid points |
| Laser vibrometer | non contact velocity measurement |
| FFT analyzer | builds the frequency response functions |
Vibration testing rule: the resonance is a measured property, not a mystery, and the mode shape is the map of the fix. Separate the excitation from the resonance, cap the amplification with damping, and stiffen or move the mass at the antinode, and the machine that shook itself apart runs quiet again.
Vibration testing and modal analysis turn the ringing of a structure into the engineering data: the hammer and the shaker make the structure speak, the accelerometers and the analyzer write down its voice, the parameter estimation extracts the modes, and the validated modal model becomes the tool of the design office, the plant and the monitoring station. The natural frequency, the damping and the mode shape, the three numbers of every mode, connect the measured vibration to the designed fix, and the engineer who reads the mode shape sees where the rib, the mass or the damper belongs before the first modification is made. The structure that was a mystery of noise and failure is now a set of well understood modes, each with its frequency, its shape and its cure, and the vibration that once shut down the machine is the same vibration that now tells the engineer exactly what to change.
