What Is Structural Health Monitoring? Listening to a Building Vibrate

1. The basic idea

Every structure has its own natural frequencies and modes of vibration. They depend on its mass and stiffness; for a single-degree-of-freedom system the natural frequency is proportional to √(k/m). Damage — cracked concrete, yielded reinforcement, a loosened connection — reduces stiffness; as stiffness falls, frequency drops and the mode shape changes.

Structural Health Monitoring (SHM) reads this chain backwards: it measures the vibration of a structure continuously, tracks how frequencies and shapes change over time, and tries to draw conclusions about the structure's condition from that change.

2. What is measured?

• Accelerometers: the most common sensors. They measure acceleration at several points; frequencies and mode shapes are extracted from the records.
• Strain gauges: measure local deformation in a member; important for fatigue and load history.
• Tilt and displacement sensors: track permanent movements such as settlement or tilting.
• Environmental sensors: temperature, humidity and wind. They do not measure damage — but, as shown below, without them damage is very hard to interpret.

For a sensor's raw signal to become a meaningful frequency it must be sampled and processed correctly. These two steps are the subject of the site's first two SHM articles.

3. Four questions, four levels

Rytter (1993) defined the goals of damage identification at four levels:

1. Existence: is there damage?
2. Location: where is it?
3. Extent: how severe is it?
4. Prognosis: what is the remaining life and safety of the structure?

Each level is much harder than the previous one. Vibration-based methods are strong at the first level; locating and quantifying damage requires many more sensors, a good model and careful interpretation.

4. The greatest challenge: the environment

A structure's frequency does not change with damage alone. Temperature, humidity, traffic load and support conditions also change it. In the well-known one-year monitoring of the Z24 bridge in Switzerland (Peeters and De Roeck, 2001), frequencies rose markedly when temperatures fell below freezing; this change was comparable in size to the effect of the damage deliberately applied afterwards.

So "the frequency dropped, therefore there is damage" can be wrong. The first question is whether the change exceeds the measurement uncertainty and the environmental variability. That question is at the heart of the site's SHM-03: when is a frequency shift evidence?

5. After an earthquake

The period right after an earthquake is when monitoring is most valuable. The most urgent question is simple: is this structure safe? For a structure monitored before the earthquake, comparing frequencies and shapes after the event with those before can give a quick first assessment. But this requires a reliable "healthy" reference recorded beforehand; a system installed after the earthquake cannot know the previous state.

Such a preliminary assessment does not replace an on-site engineering inspection; it helps decide where to inspect first.

6. SHM articles on this site

The site's seven advanced SHM articles follow, in order, the path of monitoring data from measurement to decision:

1. SHM-01 — Sampling and aliasing: the error that cannot be fixed after recording
2. SHM-02 — FFT, windowing and damping
3. SHM-03 — When is a frequency shift evidence?
4. SHM-04 — Do two independent methods find the same shape?
5. SHM-05 — Measuring torsion
6. SHM-06 — MAC, COMAC and frequency shift
7. SHM-07 — Damage, repair, strengthening