Summary: Placement, not sensor count, decides whether a tilt monitoring scheme can answer the question it was installed for. A single sensor measures rigid body rotation, and rigid body rotation causes no stress or strain inside a structure; what actually damages a building is angular distortion, the change of slope between two points. Separating the two requires at least two points at a known spacing on a common axis, and attributing movement to the right mechanism usually requires a chain. This guide sets out minimum schemes by structure type, spacing rules including the ISO 18674-3 gauge length limit, the baseline period that must precede any trigger value, and the placement mistakes that quietly invalidate a scheme.
1. What a Single Tilt Sensor Can and Cannot Tell You
A single tiltmeter answers one question well: how far has this point, on this surface, rotated away from the position it held when it was zeroed. That is a complete and correct answer to a narrow question.
It cannot tell you why. Consider a tiltmeter at the top of a 12 m bridge pier reading 0.002 degrees. Every part of that reading is correct, and it is compatible with at least three different mechanisms: a plastic hinge forming at the base of the shaft, bending distributed along the height of the pier, or the foundation settling differentially so that the whole pier rotates as a rigid body. Each of those calls for a different intervention and implies a different remaining life for the structure. One sensor cannot separate them.
The same ambiguity appears elsewhere. On a slope, a single instrument confirms that the ground is moving but says nothing about the depth of the sliding surface. On a retaining wall, it cannot distinguish bending towards the excavation from rotation at the top.
2. Rigid Rotation Versus Angular Distortion
Rigid body rotation, the entire structure tipping as one piece, causes no stress and no strain inside the structure. It matters for serviceability, for clearances and for whatever the structure leans against, but it does not crack a wall by itself.
What causes damage is angular distortion: the change of slope between two points along the structure. The distinction is written directly into the classical damage criteria. Angular distortion is defined as the distortion between two adjacent points, that is the differential settlement divided by the distance between them, minus the rigid body rotation of the building. Modern criteria follow the same logic. Boscardin and Cording showed in 1989 that horizontal ground strain also contributes, and that the more horizontal strain a building experiences, the less angular distortion it can tolerate before damage appears.
The consequence for placement is unavoidable. Isolating angular distortion requires two points at a known distance apart. A single sensor measures rotation; it cannot measure distortion, because distortion is a property of the interval between two points, not of a point.
For scale, the classic serviceability limit for ordinary structures is an angular distortion of 1/500, equivalent to 2 mm per metre, from Skempton and MacDonald’s 1956 study of settlement damage. Published thresholds for visible damage sit above that. Cracking of walls in steel or reinforced concrete framed buildings has been associated with angular distortion of roughly 3.3/1000, moderate to very severe damage in brick bearing walls with roughly 3.25/1000 in the Boscardin and Cording damage charts, and structural damage in beams and columns with roughly 6.6/1000.
3. Minimum Monitoring Schemes by Structure Type
The minimum scheme is the smallest set of positions that can separate the mechanism you are monitoring from the other mechanisms that would produce the same reading on a single sensor.
| Structure | Minimum scheme | What the extra points catch |
| Building adjacent to an excavation | One biaxial unit low on the excavation-facing facade, one at mid-height, one on the opposite side | Rotation at foundation level, racking of the upper storeys, and differential behaviour between the two sides |
| Tall building or tower | Additional units at intervals up the near facade, denser towards the base | A tilt profile with height, separating foundation rotation from deformation of the upper storeys |
| Retaining wall or excavation support | Along the wall, at anchor levels, and wherever the design expects maximum bending | Whether the wall bends towards the excavation, rotates at the top, or moves as a rigid block |
| Slope or active landslide | A chain down a borehole through the moving mass, plus surface points across the slope | The depth of the sliding surface, not only confirmation that the slope is moving |
| Bridge pier, pylon or column | A vertical chain along the shaft, denser near an expected hinge or a change of section | Flexural deformation as distinct from rigid body rotation of the foundation |
| Tunnel lining and portal | Points around the lining section, plus the portal and any structure above the alignment | Convergence of the section and rotation of the portal structure |
| Machinery, mast or temporary works | At the points where loss of level becomes a functional failure | Loss of level at the point that matters, before the asset stops working |
4. Sensor Spacing: How Far Apart, and Where to Densify
A chain is only as informative as the positions of its elements. Published practice for tiltmeter chains places elements between roughly 0.5 m and 10 m apart, selected against the spatial scale of the deformation expected rather than against how many instruments happen to be available. Where the segment between two successive sensors can be treated as straight, the same principle extends to curved geometries such as arch extrados or vault intrados.
ISO 18674-3 constrains the other end of the same problem. For in-place inclinometers it requires that the gauge length, the distance between measurement positions, does not exceed 2 m, and notes that shorter gauge lengths commonly produce better results. For portable probe inclinometers the same standard requires each increment along the measuring line to be no greater than the gauge length of the probe, and the depth measuring device to carry permanent, wear-resistant marks at that same spacing.
That last requirement is a placement rule in disguise. If the marks on the cable no longer match the gauge length of the probe, the profile you reconstruct is wrong before any sensor error is taken into account.
Uniform spacing is the simplest layout to design and rarely the most effective one, because deformation concentrates. If the design expects a hinge at the base of a pier, an anchorage level in a wall, or a sliding surface at a particular depth, elements should be dense there and can be sparser elsewhere. On a chain whose elements can be supplied in different lengths, that is a matter of specification. On a portable survey, it means choosing depth intervals deliberately rather than running a uniform step down the whole hole.
5. The Baseline You Must Log Before Any Trigger Value
Placement also fixes what can serve as a reference, and a reference only exists after the structure has been observed in its normal state.
Field practice for buildings adjacent to excavation work is to log one to two weeks before the works begin, ideally covering at least one change in weather. That interval is what reveals the amplitude of the building’s normal daily movement, the phase lag between air temperature and the structural response, and, on occupied buildings, the difference between weekdays and weekends.
The first readings should not be read as tilt data. On a sun-exposed concrete face with a daily temperature swing of 15 to 20 degrees Celsius, combined structure and sensor movement can reach 0.3 to 0.6 mm per metre peak to peak before any compensation, which means the first roughly 48 hours are better read as thermal stabilisation of a newly installed sensor than as movement of the structure.
Existing buildings also arrive with a history. A masonry building from the 1920s can carry a historic lean of 10 to 20 mm per metre, and that lean is usually not what the monitoring is about. Once installed, each sensor is levelled and zeroed against a surveyed reference so that monitoring starts from a defined position rather than an arbitrary one.
6. How Placement Defines What a Trigger Value Means
A trigger expressed as angular distortion needs two points, at a known spacing, on the same axis. That is a placement requirement before it is a threshold requirement. Change the spacing and the same physical movement produces a different angular distortion, so the trigger value and the layout have to be designed together.
Two rules follow from the baseline. First, an alert level must sit clear of the structure’s normal variation. If a facade swings 0.4 mm per metre every afternoon and the alert is set at 0.5, the system will alarm on most days, and the monitoring team will learn to ignore it. Those alarms are still logged, which is worse than not having them. Second, a cumulative trigger should always be paired with a rate trigger. The same cumulative rotation means different things at different speeds: a wall rotating slowly in step with a construction programme is behaving as designed, while the same total accumulated in the 48 hours after a dewatering pump fails is an emergency.
Sensor capability sets the floor for both. A dual-axis unit such as the ZCT2xxM-LBS-Ax-H5-460x or the ZCT330Mx-SWP-N-YKC1 publishes accuracy of 0.005 to 0.01 degrees with 0.001 degree resolution, and a 4-20 mA analogue unit such as the ZCT205M-LPS-7205 publishes 0.005 to 0.01 degrees over a range of plus or minus 5 degrees. Those figures determine how small an angular distortion a given spacing can actually resolve, which is the other half of the layout decision.
7. Common Placement Mistakes
| Mistake | What you see later | Root cause | Fix |
| One sensor on a large structure | Correct readings, but no way to attribute the movement to a mechanism | Rotation and distortion are the same single number to one sensor | Add a second point on the same axis at a known spacing |
| Every point at one elevation | Rotation confirmed, distribution with height unknown | No profile along the structure | Add units up the facade or along the shaft |
| Uniform spacing through a known weak zone | Deformation resolved everywhere except where it concentrates | Layout designed around instrument count | Densify at hinges, anchorages and expected sliding surfaces |
| A slope instrumented at a single depth | Movement confirmed, sliding surface unknown | No depth resolution | Install a chain, or step a portable probe at the gauge length |
| Triggers set before any baseline | Alarms every afternoon, then ignored | Thermal amplitude of the structure not measured | Log one to two weeks first, then set alerts above that amplitude |
| Sensor fixed to cladding or a non-structural element | Readings follow sun and wind rather than the structure | Reference not tied to the element being monitored | Mount on the structural element the movement is attributed to |
| Elements at unequal, unrecorded spacing | Displacement profile shifted by a constant offset | Spacing assumption in the processing is wrong | Record actual spacing and feed it into the profile calculation |
8. A Five-Step Placement Planning Workflow
| Step | Question to answer | Output |
| 1. Name the mechanism | Which failure mode is this scheme meant to catch? | A mechanism, not a structure name |
| 2. Name the smallest point set | Which positions separate that mechanism from the others? | A minimum scheme for the structure in front of you |
| 3. Choose the geometry | Do the points share an axis, and is a chain needed for depth or height? | A point list, or a chain specification |
| 4. Set spacing against the deformation scale | Where is deformation expected to concentrate? | Non-uniform positions, dense where it matters |
| 5. Log the baseline, then set triggers | What does normal look like on this structure? | Alert, alarm and rate levels with margin above normal |
Frequently Asked Questions
- How many tilt sensors do I need to monitor a building next to an excavation? Three is the practical minimum: one low on the excavation-facing facade, one at mid-height, and one on the opposite side to show whether the building is moving as a unit or distorting internally. Taller buildings need additional units up the near facade so that a profile with height can separate foundation rotation from deformation of the upper storeys.
- Can a single tilt sensor detect a slope failure? It can confirm that the ground is moving, but not where the movement is happening. Locating the sliding surface requires a chain of sensors down the borehole, or a portable probe surveyed at intervals no greater than its gauge length, because the depth of the shear zone is what the stabilisation design depends on.
- What spacing should I use between tilt sensors in a chain? Published practice spans roughly 0.5 m to 10 m, chosen against the spatial scale of the expected deformation rather than against the number of instruments available. ISO 18674-3 caps the gauge length of in-place inclinometer elements at 2 m and notes that shorter lengths commonly give better results. Spacing should be tighter wherever deformation is expected to concentrate.
- Do I need a baseline before setting any alarm level? Yes. One to two weeks of logging before the works start is standard field practice, ideally covering a change in weather, so that the structure’s normal daily amplitude and its phase lag against temperature are known. The alert level must then sit clear of that amplitude. Otherwise the scheme produces regular false alarms and the team stops responding to them, which defeats the purpose of installing the system.
TAG: