Summary: Ground that freezes moves every winter, and a tilt sensor sees only part of that movement. Heave and thaw settlement are vertical, and a footing that lifts or drops as a whole produces no rotation at all, so the reading contains only the differential part, which is also the part that cracks walls and distorts track. That differential is not small: 10 millimetres across a 5 metre footing is 0.115 deg, twenty three times the accuracy class of a high precision tiltmeter. The harder problem is that this particular cycle does not cancel, because frozen ground does not retrace its path, so the seasonal component must be compared between years rather than subtracted. And in the same cold week the instrument has its own winter, in which a cell rated to minus 40 degrees Celsius may still be unable to deliver the transmit pulse.
1. Why the Usual Rule Fails in Freezing Ground
Elsewhere in this series the seasonal cycle is treated as a nuisance. A structure that expands in summer and contracts in winter, or a reservoir whose level rises and falls, produces a movement that is large and completely reversible: the reading goes out and comes back, and the only information is in what does not come back. The standard method follows from that, which is to fit the cyclic component, subtract it, and keep the residual.
That method carries a condition that is easy to miss. Subtracting a cycle is only legitimate when the mechanism is elastic and therefore retraces its own path. Where the mechanism changes state, the path out and the path back are different, the cycle leaves a residue in the same direction every time, and the residual is no longer the only information in the record. The shape of the cycle is information too.
Ground that freezes is the clearest case of a state change in structural monitoring. Water leaves the soil and becomes ice, the ice moves soil and structure upward, and in spring the ice becomes water again in a place where it cannot drain. This guide works through what that does to an angle measurement, how large the resulting rotation is in the units an engineer uses, and why the same cold weather that moves the ground also puts the instrument itself at risk of failing to report.
One boundary before starting, because three nearby questions belong to other guides in this series. Thermal drift of the sensor, meaning the change in a tilt reading caused by the sensor’s own temperature, is a different problem and is treated separately in this series. Battery life, meaning how many years a cell lasts at a given reporting interval, is also a different problem. And a third boundary matters more than either: the two quantities in this article, vertical heave and angular rotation, are not the same measurement, and a tilt sensor produces only the second.
2. What Freezing Ground Does: Three Conditions and One Mechanism
Frost heave is often described as water expanding as it freezes. That description is incomplete in a way that matters. Water does expand by about 9 percent when it turns to ice, but that expansion is limited to whatever pore water happens to freeze where it already sits, and much of it is absorbed into void space rather than lifting the soil skeleton. The movement actually observed is produced by water that was not in the frozen zone at all: it migrates upward to the freezing front, freezes there, and the ice that forms is new volume added to the soil. The process is called ice segregation, and it is why a heave of several centimetres is possible in ground that contains nothing like that much water in the zone that froze.
Segregation works like this. As the freezing front advances, the coarse pores freeze first and the water in them is consumed. That leaves the remaining unfrozen water at a lower chemical potential, which appears as suction, and the suction draws water from the warmer soil below toward the freezing front. The arriving water freezes onto the existing ice, thickening it into a distinct horizontal layer called an ice lens, and the lens grows in the direction of heat loss, which is upward. The heave of the surface is at least as much as the combined thickness of those lenses, and published pavement and foundation literature puts that at several centimetres in a frost susceptible soil with a good water supply.
Two numbers make the mechanism concrete. Capillary rise in a silty soil, which is the soil most prone to this, can reach 2 to 3 metres, which is why lowering a water table is one of the standard remedies: if the water cannot reach the freezing front, there is nothing to segregate. And the suction developed at the front has been reported above 100 kilopascals, which is the force that pulls water upward against gravity through an otherwise static soil.
Three conditions have to hold at the same time for any of this to happen, and removing any one of them removes the problem. That triple requirement is what makes frost heave a manageable design problem rather than an unavoidable one.
| Condition | What it means in the ground | How it is removed |
| A frost susceptible soil | A soil containing enough fine particles to sustain capillary flow toward the freezing front | Replace the material with clean granular fill, or place a capillary break above the water table |
| Freezing temperatures that penetrate the soil | Not cold air alone, but a frost front that descends below the level of the foundation or the monitored element | Insulate, or found the structure below the depth the frost reaches |
| A supply of water | A water table within capillary reach of the freezing front, or infiltration feeding the frozen zone | Lower or intercept the water table, or drain the surface so that water does not accumulate |
The size criterion for the first condition comes from a rule of thumb published by Casagrande in 1932 that is still quoted in foundation and pavement engineering. Under natural freezing conditions and with a sufficient water supply, considerable ice segregation should be expected in non-uniform soils containing more than 3 percent of grains smaller than 0.02 millimetres, and in very uniform soils containing more than 10 percent of grains smaller than the same size. Below 1 percent, no ice segregation was observed even with the groundwater level as high as the frost line. The test is a hydrometer analysis, so it is ordinary soil laboratory work and not a research procedure.
Note which soils the criterion points to. Silt is the worst case, and the reason is worth stating because it is counterintuitive. Clay has more fines and higher suction than silt, but its permeability is so low that very little water can travel to the freezing front during the weeks the front is advancing. Clean sand and gravel have the permeability but not the capillary reach, so water drains away instead of rising. Silt sits between the two, with enough capillary pull to lift water and enough permeability to deliver it, which is why the word silt appears in almost every frost heave case record.
3. The Freezing Index and the Frost Depth, and Why a Deeper Freeze Is Not More Heave
The severity of a winter, for engineering purposes, is not its lowest temperature but its cumulative cold. The freezing index is the sum of the number of degrees by which the daily mean air temperature falls below freezing, added up over the freezing season. A winter with a few intensely cold weeks and a winter with a long moderately cold stretch can have the same index, and as far as the ground is concerned they are the same winter.
The frost depth follows from the index. ISO 13793, the international standard for the thermal design of foundations to avoid frost heave, gives the greatest depth of frost penetration in undisturbed ground as the square root of a fraction. The top of the fraction is 7200 multiplied by the design freezing index multiplied by the thermal conductivity of frozen soil. The bottom is the sum of two terms: the latent heat of freezing of the water in the soil per unit volume, and the heat capacity of the unfrozen soil per unit volume multiplied by the annual mean air temperature. The index is expressed in kelvin hours, and the standard’s recommended values for a worst case of bare ground over granular soil are a thermal conductivity of 2.5 watts per metre kelvin, a latent heat of 150 million joules per cubic metre and a heat capacity of 3 million joules per cubic metre kelvin. The standard expects national maps or tables to supply the depth directly for many locations, and provides the formula for everywhere else.
Running that formula at an annual mean air temperature of 5 degrees Celsius gives the depths in the table. The absolute values depend on the climate chosen, but the shape of the relationship does not, and the shape is the useful part.
| Freezing index | Computed frost depth in undisturbed ground | Relative depth |
| 100 degree days | 0.51 metres | 0.71 |
| 200 degree days | 0.72 metres | 1.00 |
| 300 degree days | 0.89 metres | 1.22 |
| 500 degree days | 1.14 metres | 1.58 |
| 800 degree days | 1.45 metres | 2.00 |
| 1,200 degree days | 1.77 metres | 2.45 |
This is a square root relationship, and the consequence is worth saying out loud because it is used badly in procurement conversations. Doubling the freezing index raises the frost depth by about 41 percent, not by 100 percent. Quadrupling it doubles the depth. A site that is twice as cold as another is not twice as deep frozen, and a specification written for one winter is not invalidated by a harder one in proportion to the temperature difference.
A simpler published form makes the same point and adds a surface condition. The ASCE 32-01 method writes the depth as a soil coefficient multiplied by the square root of the freezing index, with coefficients of about 67, 57 and 50 millimetres per root degree day for granular, medium grained and fine grained soils. At an index of 500 degree days those three coefficients give 1.50, 1.28 and 1.12 metres, which straddle the 1.14 metres from the ISO method, so the two published methods land within about 30 percent of each other on the same climate and are useful as mutual checks. More interesting is the surface factor that the FHWA guidance adds on top: 1.0 for standard ground and 1.3 for a road kept clear of snow. A ploughed road exchanges heat with the cold air directly while a snow covered field is insulated by the snow, and because the depth varies as the square root, a factor of 1.3 in the index is worth 14 percent of extra depth. Removing snow from a road, or from a monitored slab, changes what the ground underneath experiences.
Now the part that matters most for monitoring, and it is an inversion. A deeper frost is not a larger heave. The thickness of an ice lens depends on how slowly the freezing front advances and on how much water can migrate to it before the front passes, and the two independent published statements of the mechanism agree that a quick freeze does not have sufficient time to form lenses of any significant size. A long, slow, moderately cold winter can therefore heave a foundation more than a short severe one, because the slow front gives the water time to arrive. The freezing index tells you how deep the ground freezes. It does not tell you how much the ground will lift, and it is the lift that bends the structure.
One more detail from the same standard belongs in this section because it is, in effect, a monitoring instruction. The standard notes that heat from a building raises the ground temperature less at the corners than along the sides, because a corner has two exposed faces where a side has one. It therefore requires additional measures at corners, either deeper foundations or extra insulation, and it treats the corner case as a separate design situation. A standard that expects the corners of a building to experience a different ground thermal regime from the sides is telling the monitoring engineer something precise: the differential movement that a tilt sensor is best at detecting is not a random irregularity, it is a predicted feature of the site.
4. What a Tilt Sensor Sees: Rotation, Not Lift
A tilt sensor measures the inclination of its own mounting face against gravity. It reports an angle, and an angle is a ratio of two lengths, so the reading is produced by the difference between two points rather than by the movement of either one. In freezing ground that distinction decides everything.
If a footing heaves uniformly, by 25 millimetres over its whole length, every point of it rises by the same amount and the face it presents to the sensor is still horizontal. The angle has not changed, so the sensor reports nothing at all, and no adjustment of its resolution changes that. On a foundation that lifts as a rigid whole, the instrument is blind, and its reading contains only the difference between the movements at the two ends of whatever span its mounting rigidly bridges.
That difference converts straight into an angle. Over a mounting baseline of length L, a differential vertical movement d produces a rotation of d divided by L, which for these sizes is the same as the angle whose tangent is d over L to within a fraction of a percent. The table gives the arithmetic for the differential movements that frost actually produces.
| Differential movement | Across 1 metre | Across 5 metres | Across 10 metres |
| 2 millimetres | 0.115 deg | 0.023 deg | 0.011 deg |
| 5 millimetres | 0.287 deg | 0.057 deg | 0.029 deg |
| 10 millimetres | 0.573 deg | 0.115 deg | 0.057 deg |
| 20 millimetres | 1.146 deg | 0.229 deg | 0.115 deg |
| 30 millimetres | 1.718 deg | 0.344 deg | 0.172 deg |
Three points follow from that table. The first is about capability. Ten millimetres of differential across a 5 metre footing is 0.115 deg, which is twenty three times the 0.005 deg accuracy class of a high precision tiltmeter, and against the 0.001 deg resolution class of the same instrument it is 115 counts. The second is about range, which is not under pressure either: the largest entry in the table, read across the shortest baseline, is 1.718 deg, about a third of the range of a plus or minus 5 degree sensor, and the same differential read across a 5 metre footing uses only 7 percent of it. In freezing ground the range decision is driven by the mounting baseline rather than by the climate.
The third point is the reason the blindness does not matter. A uniform heave does no damage. Differential heave is what cracks a masonry wall, opens a joint between two elements, rotates a pier away from the element it supports, and destroys the geometry of a track. The quantity that damages the structure and the quantity the instrument is best at measuring are the same quantity, and the one it cannot measure is the one that is harmless. That is the opposite of the usual relationship between an instrument and its subject, and it is worth stating plainly because it changes what the monitoring plan should record: the sensor’s mounting baseline is not a detail of the installation, it is the scale factor of the whole measurement.
One honest limitation follows from the same arithmetic in the other direction. A sensor mounted on a single small bracket, bridging perhaps 100 millimetres of surface, is measuring a differential across 100 millimetres, so a 0.5 millimetre edge lift reads as 0.287 deg while the same 0.5 millimetres spread over a 5 metre footing reads as 0.006 deg. The same ground movement produces two very different numbers depending on how much of the structure the mounting spans. Any comparison between two monitoring points, or between two campaigns, has to be made at the same baseline or it is not a comparison.
5. The Hysteresis Test: Why This Cycle Does Not Cancel
In an elastic material a cycle is a closed loop. Load it and it deforms, unload it and it returns along the same path to the same place, and the area inside the loop is zero, which is another way of saying that the cycle produced no permanent change and therefore carries no information about damage. That is the assumption that justifies subtracting a seasonal component, and it is a good assumption for a steel or concrete structure responding to temperature.
Frozen ground violates it at every step. On the way into winter, water migrates toward the freezing front and freezes there, so the soil that ends up frozen contains more water than the soil that started, and the ground surface rises. On the way out, the ice melts and releases that water into a soil whose permeability is low enough that it cannot drain as fast as it appears. The soil is temporarily saturated and weak, it consolidates under its own weight and the weight of whatever is standing on it, and the surface comes to rest at a level that is not necessarily the level it started from. The path down is not the path up. The loop has an area, and the area is the part that matters.
The practical consequence is a change in data treatment, not a change in instrumentation. Do not fit the season and subtract it. Compare like with like across cycles: this year’s winter maximum against last year’s winter maximum, and this year’s post thaw minimum against last year’s. Under the elastic assumption the two extremes carry the same information, because the cycle is symmetric. Here they do not, because the two halves of the cycle are produced by different mechanisms.
Comparing the two extremes separately produces four combinations, and each points at a different cause. This is the diagnostic that replaces the subtraction.
| Winter maxima | Thaw minima | What the pattern points to |
| Rising season by season | Steady | The heave amplitude is growing, so the water supply reaching the freezing front is increasing or drainage has deteriorated |
| Steady | Falling season by season | Irreversible settlement is accumulating, which is the signature of thaw consolidation rather than of frost |
| Both drifting the same way by the same amount | Translating together | The whole cycle has moved, so this is a structural or foundation trend and not a frost effect at all |
| Shrinking | Rising | The amplitude is falling, which usually means drainage has improved or the ice rich layer has already been consumed |
The extreme case of the second row is worth a number, because it is what happens when the ground ice is not seasonal but permanent. A Geological Survey of Canada network of thaw tubes in the Mackenzie Delta, seventeen sites measured between 1991 and 2016, found thaw penetration increasing at ten sites at a median rate of 0.5 centimetres per year, and ground surface subsidence at ten sites at a median rate of 0.4 centimetres per year. Over the twenty five years of the record between about 5 and 38 centimetres of permafrost thawed and about 5 to 20 centimetres of excess ground ice melted. The most instructive single result is a site near Inuvik, where the active layer thickened by 19 centimetres between 1994 and 2016 while thaw penetration increased by about 40 centimetres and the surface subsided by 20 centimetres. In ice rich ground the thaw penetration can therefore increase by more than twice as much as the active layer thickness, which is why probing the active layer alone cannot detect the degradation. A separate pan Arctic synthesis of the same phenomenon reports thaw subsidence rates up to 2 centimetres per year where ice content is low and more than 3 centimetres per year in ice rich permafrost, with higher rates where there has been human activity or recent wildfire.
Put that in the units of this series. Three centimetres per year, if it were fully differential across a 10 metre baseline, is 0.017 deg per year, and across a 1 metre baseline it is 0.172 deg per year. The angular numbers are small, and that is an honest finding rather than a problem: these are slow millimetre scale movements and the geotechnical engineer will state them in millimetres. What matters is that the angle channel can resolve them with margin. The 0.001 deg resolution class of a monitoring tiltmeter, over a 10 metre baseline, is 0.17 millimetres, so a 30 millimetre per year differential is about 170 resolution steps in a year. The limit on detecting it is not the instrument, it is how many years of comparable record exist.
6. Spring Is When the Damage Happens
The winter heave is large and visible and mostly harmless, and the event that actually moves a structure permanently happens at the thaw. The mechanism has a name in pavement and foundation engineering, thaw weakening, and the description is consistent across sources. As the ice lenses melt, the released water saturates the soil faster than it can drain, the soil loses stiffness and bearing capacity, and loads that would do nothing to frozen ground or to a drained subgrade can be damaging at that moment.
This reorganises the monitoring plan in three ways. The first is which extreme to watch. If the structure can move permanently, it will do so in the weeks after the thaw rather than at the coldest point of the winter, so the post thaw minimum is the number to compare year on year and the season to alarm on is the thaw. The second is when to be cautious about loading. A construction programme, a heavy vehicle movement or an excavation that would be routine in a drained season is not routine on a soil that has just lost its ice. And the third is that a short intense thaw and a long slow one behave differently, because published pavement research describes melting as proceeding from the surface downward, from the base upward, or from both at once depending on the surface temperature, and a thaw that starts at the surface releases its water into a profile whose lower part is still frozen. Water arriving above an impermeable frozen layer has nowhere to go, which is why the softening can be more severe than the quantity of meltwater alone would suggest.
A useful consequence for anyone interpreting a record is that the two halves of the year have different shapes as well as different signs. The heave is gradual, driven by the progressive advance of the freezing front, and it reverses when the front retreats. The thaw settlement is relatively abrupt, driven by the loss of ice and the consolidation that follows it, and it does not reverse. A single channel recording both will show a slow rise and a faster drop, and the asymmetry in the shape is the first indication that the second mechanism is present.
7. The Instrument’s Own Winter: A Rating Is Not a Capability
Everything above is about the structure. The instrument has a winter of its own, and in the same week, which makes the two easy to confuse. The first specification to get right is the operating temperature range, and products in this series do publish it: the ZCT-IOTH-WN-2x-JLA8 wireless tiltmeter is rated from minus 40 to plus 85 degrees Celsius, and so is the ZCT330M-SWP-L-CN470 used on a railway tunnel project in Qinghai Province at an altitude above 3,000 metres.
An operating range guarantees that the sensor functions. It does not guarantee that the sensor can transmit, because a radio transmission is not a continuous draw but a pulse, and a pulse needs current. The same datasheet gives the two numbers that describe the gap: a peak current of 300 to 1,000 milliamperes, against a quiescent current of 210 to 220 microamperes with the alarm path enabled and 5 to 10 microamperes with it cancelled. A transmission therefore draws roughly four to five orders of magnitude more current than standby, and the cell has to deliver it in a burst, at whatever temperature the enclosure happens to be.
Cells in this class do not deliver the same pulse at every temperature. A published cell level pulse test on a 3.6 volt lithium thionyl chloride cell, which is the same chemistry the tiltmeter above uses, recorded the rest voltage, the voltage after a 10 second pulse and the recovery voltage at three temperatures. The result is in the table.
| Cell temperature | Rest voltage | Lowest voltage during a 10 second pulse | Voltage after recovery |
| Plus 25 degrees Celsius | 3.65 volts | 3.50 volts | 3.64 volts |
| Minus 20 degrees Celsius | 3.62 volts | 3.32 volts | 3.58 volts |
| Minus 55 degrees Celsius | 3.60 volts | 2.80 volts | 3.20 volts |
The sag under load grows from 0.15 volts to 0.30 volts to 0.80 volts, which is a factor of about 2.7 between the two cold rows even though the rest voltage barely moves. The rest voltage is the number a user checks, and it is the number that stays healthy. The number that decides whether the transmission succeeds is the loaded voltage, and it is not on the label.
Put the second row and the third row against a module that needs 3.00 volts to operate, which is the order of magnitude for a cellular module. At minus 20 degrees Celsius the pulse bottoms out at 3.32 volts, leaving 0.32 volts of margin, and the transmission completes. At minus 55 the pulse bottoms out at 2.80 volts, which is 0.20 volts below the threshold, and the same cell supplying the same load cannot make the transmission. Nothing in the chemistry changed between the two rows; the temperature did.
The mechanism has a name and a consequence that matters for how a monitoring node is built. A layer of lithium chloride forms naturally on the lithium anode and protects the cell from self discharge. In the cold that layer becomes thicker and more resistive, so the first significant load applied after a quiet period has to break it down, and if the current is not sufficient the voltage collapses instead. The effect is called voltage delay, and it means the transmission most likely to fail is the first one after the deepest cold, which is precisely the transmission that carries the alarm. A cell design with a thinner passivation layer, a capacitor that stores charge between transmissions, or a small wake up pulse issued before the real one are the three routes that published battery engineering gives for the same problem.
Two more consequences follow from the same physics, and both are usually missed. The first is that usable capacity is not a property of the cell alone. A device whose module resets at 3.00 volts loses far more of its cold capacity than one that keeps working down to 2.00 volts, because the chemical energy is still present but becomes inaccessible once the loaded voltage crosses the threshold. The second is on the datasheet already and is easy to overlook: the same tiltmeter datasheet recommends a stocking temperature of plus 25 to plus 30 degrees Celsius when the cell is fitted. A node that spends a winter in an unheated site store, or a summer in a shipping container, has been stored outside that recommendation before it was ever installed. The general rule to carry away is that the cold limit of a battery powered instrument is a property of the cell and the load together, and the figure to check is the loaded voltage at the worst expected temperature against the cut off voltage of the radio.
8. Three Candidates, One Channel: Attributing a Winter Anomaly
The standard method for separating an instrument artefact from real movement is to log temperature beside the angle and use the relationship between them. It works because the two have different shapes: a sensor zero that moves with temperature produces a daily cycle that tracks the temperature record, while a structural trend is slow and monotonic.
In a cold week that method stops working, and the reason is structural rather than statistical. Three independent things are now all functions of the same temperature. The ground is moving because it is freezing or thawing. The cell may be failing to deliver its pulse because it is cold. And the sensor’s own zero is moving with temperature, at a published typical rate of plus or minus 0.001 deg per degree Celsius for the tiltmeter quoted above, which over a 60 degree Celsius swing is 0.06 deg, larger than the 0.05 deg zero point deviation the same datasheet publishes and twelve times its 0.005 deg accuracy class. The temperature channel cannot separate causes that are themselves functions of temperature.
Attribution therefore has to come from the shape of the anomaly and from the evidence recorded beside it, not from a correlation with temperature.
| What the record shows | Most likely cause | What settles it |
| Readings missing for a run of days, with no change before or after | The transmission, not the ground | The sensor’s own recorded temperature and battery voltage at the last successful report |
| A smooth excursion over hours to days that returns fully | Either movement at the freezing front or a zero shift | Two sensors on the same structure, and a regression of angle against the recorded temperature |
| A step that persists and is not recovered | Real movement of the mounting or the element | The next reading, since a step does not go away, and a physical inspection of the mounting |
| Readings arriving late, doubled or out of sequence | Link quality or retries rather than the measurement | The gateway log rather than the sensor record |
The practical conclusion is a commissioning requirement rather than an analytical one. A monitoring system that stores only the angle, and discards the sensor’s internal temperature and its supply voltage, has thrown away the two channels that would answer this question, and it has done so at no saving, because both values are already measured inside the instrument in order to produce the angle. Ask for them to be logged before the winter, because the anomaly cannot be reprocessed later if the evidence was never recorded.
9. Placement, the Baseline, and What This Cannot Tell You
Where the sensor goes decides which quantity the record contains, and in freezing ground the choice is between two genuinely different measurements. The ISO foundation standard gives the rule for the structure itself: if the foundation depth is at least the maximum frost depth in undisturbed snow free ground, the foundation is adequately protected against frost heave and neither edge insulation nor ground insulation is required, and a well drained layer that is not susceptible to frost may be counted toward that depth. Where the foundation is shallower than the frost depth, the same standard offers insulation options in place of depth.
That rule translates directly into a monitoring decision. A sensor mounted on an exposed element at or above the frost line measures the heave of the ground, because it is riding on the part that moves. A sensor mounted on a structural element founded below the frost line measures the structure’s own rotation, because the ground under it is thermally stable. Both are legitimate, they answer different questions, and a monitoring plan that mixes them at different points without saying so produces a record in which two series are not comparable. The mounting baseline adds a second layer to the same problem, in the way section 4 describes.
Two commissioning habits follow. The first is at corners. The foundation standard expects greater movement at the corners of a heated building than along its sides, because heat reaches the ground less at a corner, so a scheme that places one sensor at the midpoint of each elevation is monitoring the least demanding location. The second is the season of the baseline. A baseline captured in late summer and compared with a reading taken in February is not a measurement of movement, it is a difference between two states of the ground, and no amount of subsequent correction recovers the missing reference. Either record the baseline in the same season as the readings it will be compared with, or record enough of the first full year to establish both extremes.
Finally the limits, stated plainly because they are the reason this remains one channel among several. An angle channel cannot give the depth of the freezing front, which needs a temperature string in the ground. It cannot give the water table or the pore pressure, which are the two quantities that decide whether the mechanism is active at all, and which are measured with piezometers. It cannot give the total vertical movement of a footing, only the differential part, so a settlement survey answers a different question from a tilt record and the two do not substitute for each other. And it cannot decide whether a given differential heave is acceptable, because that limit belongs to the design of the structure and to the standard governing it, not to the instrument.
What an angle channel does well in freezing ground, and better than most alternatives, is run unattended for years at many points at once, which is what the year on year comparison in the hysteresis test requires. The signal is not a maximum, it is a trend across winters, and a trend across winters needs instruments that are still reporting in year five.
Frequently Asked Questions
- My tilt readings rise every winter and come back every spring. Does that mean the sensor is faulty? No, and it does not mean the structure is damaged either. It is the expected signature of a foundation on ground that freezes, because differential heave and thaw settlement both produce a real rotation that reverses with the season. A sensor fault would look different: it would track the daily temperature cycle rather than the seasonal one, and it would not be reproduced by a second sensor on the same structure. The number worth watching is not this winter’s maximum but whether this winter’s maximum and this spring’s minimum sit where the previous year’s did, which is the comparison described in section 5.
- Can a tilt sensor replace a settlement survey or a ground temperature string? No, because each of the three measures a different quantity. A tilt sensor reports rotation, so it sees the differential part of a vertical movement and is blind to uniform heave or uniform settlement, which a survey measures directly. A ground temperature string gives the depth of the freezing front, which an angle channel cannot infer because the two are related only through the soil’s thermal properties. What the angle channel adds is coverage and continuity: it can be installed at many points and it reports through winters when nobody visits the site, so it tells you which locations to survey rather than replacing the survey.
- Why did my wireless sensor stop reporting in the coldest week and then recover by itself? Because the transmission, not the sensor, was the part that failed. The datasheet operating range may be minus 40 degrees Celsius, but the radio needs a pulse of 300 to 1,000 milliamperes, and a cold lithium thionyl chloride cell sags far more under that pulse than its healthy rest voltage suggests. A published cell test shows the sag growing from 0.30 volts at minus 20 degrees Celsius to 0.80 volts at minus 55, which against a module needing 3.00 volts is the difference between working and not working. Recovery on its own is the confirmation, because a cell warms up and the passivation layer on its anode is broken down by the first successful load.
- How deep does a foundation have to be before frost heave stops mattering to the monitoring? The ISO foundation standard answers this directly. If the foundation depth reaches the maximum frost depth in undisturbed snow free ground, the foundation is considered adequately protected against frost heave and no insulation is required; where the freezing index is below 2,000 kelvin hours, the standard notes that a depth of 0.45 metres or more satisfies that condition. That depth is climate specific, and the same standard provides the formula to compute it from the freezing index and the soil’s thermal properties. For a monitoring plan, the practical rule is that any sensor mounted above that depth is measuring the ground’s movement and any sensor mounted on structure below it is measuring the structure’s, and the record should say which one each point is.
TAG: