Summary: The allowable settlement of a flat-bottomed steel storage tank is not one number. The published limits run from about 1 part in 1,000 for the differential settlement around the periphery of a new tank to about 1 part in 25 for the sag of the bottom plate, so the family spans a factor of forty, and the reason is that each limit is protecting a different component rather than describing the ground. Two consequences follow. A criterion written as a millimetre allowance over a stated arc is a slope, so the same measured tilt passes or fails depending only on the length it is measured over, and a tilt sensor’s mounting base plays the part that the survey spacing plays. And the differential settlement that stops a first fill is about three and a half times tighter than the structural allowance at the same arc, which means a trigger and an acceptance limit are not the same kind of number.
1. Why a Tank Has Three Settlements and Only One Problem
A flat-bottomed steel tank standing on a ring foundation can settle in three geometrically distinct ways, and the tank standards keep them in separate categories because their consequences are different. Uniform settlement is the whole tank descending as a rigid body, with the shell and the bottom plate keeping their shape. Planar tilt is the tank leaning, so that one side of the shell base is lower than the other, again with no distortion of the tank itself. Out-of-plane settlement is everything else: a local depression in the bottom plate, a bulge or a warping where the shell meets the bottom, or a dishing of the bottom towards the centre.
Only the third of those strains the tank. Uniform settlement and planar tilt are rigid body motions, and a rigid body motion does no work on the structure, so neither of them puts additional stress into the shell or the bottom plate. What they do damage is everything attached to the tank: a rigid body motion changes nozzle elevations, adds strain to the connected pipework, and alters the drainage falls on the pad. That is why a tank can pass a structural assessment with a lean that would be alarming on a building, and why the standards give the two rigid categories their own limits rather than folding them into one number.
The tank standards are explicit about what happens when the movement stops being rigid. API 653 lists the consequences of excessive planar tilt as malfunction of floating roof seals, binding of a floating roof, problems with connecting pipes, problems with surface water drainage from the tank pad, tank shell buckling, and overstress of shell or bottom plates. The list is worth reading slowly, because the first three are serviceability consequences and the last three are structural, and the structural ones arrive indirectly, through the shell having to deflect to follow a base that is no longer a plane.
The mechanism that makes out-of-plane settlement the dangerous category is the shell-to-bottom connection. The shell is a thin cylinder stiffened by its own curvature, and the bottom plate is a membrane that carries the product’s weight in tension, so the two elements respond to the same ground movement in completely different ways. When the shell bridges over a locally soft spot while the bottom plate follows the ground down, the joint between them has to absorb the whole difference, and the published description of that failure, from Marr and co-authors in 1982 and adopted into the assessment annexes since, is a rupture of the shell-to-bottom connection rather than a buckle of the shell.
State the same thing as a design rule and it explains where a tilt channel is aimed. A tank’s shell is damaged by the part of the settlement it cannot follow as a rigid body, and the quantity that describes how hard it is being asked to bend is the change of slope of its base. A levelling survey reports elevations, from which a slope has to be recovered by differencing two readings and dividing by their spacing. A tilt sensor reports a slope directly. That is not an argument that one replaces the other, and sections 5 and 9 show why not, but it does explain why the two instruments end up quoting the same criteria in different units.
2. The Limits as Slopes
The single most useful thing a tank engineer can do with the settlement clauses is to put them all into one unit. Almost every published limit is already a ratio or a length per unit length, and the few that are written as a displacement become a ratio as soon as the span they are measured over is stated. Converted that way, the family looks like this.
| Standard | Criterion | Published value and what it governs | Equivalent slope |
| API 625 | Peripheral differential (TS) | Less than 1/1000 around the periphery, a design stage prediction for a new tank | 0.057 deg, or 1.0 mm per m |
| ACI 376 | Perimeter footing settlement | Less than 1/500 of the ring footing under the shell, taken as the lesser of this and the planar tilt limit | 0.115 deg, or 2.0 mm per m |
| PIP STE03020 | Circumferential differential | Less than L over 450 around the circumference of a tank in service | 0.127 deg, or 2.2 mm per m |
| API 650 | First hydrotest reporting trigger | 13 mm per 10 m of circumference, the point at which filling stops and the readings go to the purchaser | 0.074 deg, or 1.3 mm per m |
| API 653 Annex B | Planar tilt | Less than 1/120, the whole tank leaning as a rigid body in service | 0.477 deg, or 8.3 mm per m |
| API 653 Annex B | Out-of-plane shell settlement | 11 L squared Y divided by twice E H, capped at 101.6 mm, for local distortion of the shell base over an arc L | 0.263 deg at L equal to 10 m |
| EEMUA 159 | Out-of-verticality at the shell top | A percentage of shell height; one published secondary source quotes 0.5 percent and another quotes 1 percent | 0.29 deg or 0.57 deg |
| EEMUA 159 and PIP STE03020 | Centre to edge sagging | 200 to 400 mm for a tank of about 20 m diameter, protecting the flexible bottom plate | 1.15 deg to 2.29 deg across a 10 m radius |
Read down the last column and the first finding of this article appears. The numbers that a tank owner will meet under the single heading of acceptable settlement span from 0.057 deg to 2.29 deg, a factor of forty. If the bottom plate is set aside and only the criteria that protect the shell and its footing are counted, the span is still 0.057 deg to 0.477 deg, a factor of eight.
A discrepancy in the row for EEMUA 159 is worth keeping rather than smoothing over, because it illustrates the problem. The document certainly addresses out-of-verticality at the top of a shell, and its publisher lists planar tilt among the topics the current edition treats. Two secondary sources, both published in journals rather than in marketing material, state the numeric limit as 0.5 percent of shell height and as 1 percent of shell height respectively. Those are 0.29 deg and 0.57 deg, a factor of two apart, and the primary document is the only place that settles which applies. That is a general rule for this subject: secondary summaries of tank settlement criteria propagate their own errors, so the number that governs a decision should be read out of the standard itself.
The other thing to notice in the table is the column headed what it governs. Each of these figures is written for a component rather than for the ground it stands on, and in three cases for a specific event rather than a condition: the API 625 figure is a prediction made at design stage, the API 650 figure is a reporting trigger during the first fill, and the API 653 figures are assessments made on a tank that is already in service. Section 6 takes the difference between a trigger and a limit and shows how far apart the two can sit on the same tank.
3. Why the Same Ground Movement Gets Four Different Answers
The length of that ladder is not arbitrary and it is not bureaucratic. It follows from the tank being, structurally, two different things bolted together and asked to sit on the same ground: a thin cylindrical shell, which is stiff against local distortion, and a flat bottom plate, which is a membrane and can sag a considerable distance without being distressed at all. A subsidence bowl under a tank is therefore read as severe by whichever criterion protects the shell and as negligible by whichever criterion protects the bottom.
| Element | How it carries load | What it tolerates | The limit that protects it |
| The shell | A thin cylinder in hoop tension from the product pressure, stiffened by its own curvature | Very little local change of curvature, because bending a thin wall produces high surface stress | The 1/120 planar tilt figure and the out-of-plane allowance over an arc of 10 m |
| The bottom plate | A membrane, carrying the product’s weight in tension rather than in bending | A large centre-to-edge sag, because a membrane has no bending stiffness to overload | 200 to 400 mm for a tank of about 20 m diameter, and the R over 240 figure at design stage |
| The shell-to-bottom joint | The junction between a curved stiff element and a flat flexible one | Nothing locally, because the whole difference between the two deflections arrives at this one weld | Not a slope but a change of slope, which is why the transition has to be gradual |
| The ring footing | Spreads the shell’s line load into the ground and fixes the drainage falls | Uniform settlement, since the footing simply descends with the ground and keeps its shape | The 50 mm uniform settlement reporting trigger, set for piping strain and drainage rather than shell stress |
The second row contains a criterion that deserves a second look, because it is a slope in disguise. API 625 limits the predicted differential settlement between the edge of the tank and its centre to R over 240, where R is the radius. Divide the allowance by the distance it is measured over and the radius cancels: R over 240 across a distance R is simply 1 part in 240, whatever the size of the tank. A 20 m diameter tank is allowed 42 mm at the centre and an 80 m diameter tank is allowed 167 mm, but both of them are the same 0.239 deg of sag. A criterion written correctly as a length is quite often a constant ratio that the author has dressed as a length, and putting it back into slope form is the only way to see which.
The same trick applies to the local depression allowance in API 653 Annex B, which permits a bulge or a depression in the bottom plate of up to 30.8 times the radius of the affected area in millimetres. A depression whose affected radius is 2 m is allowed 62 mm and one whose radius is 5 m is allowed 154 mm, and here the scaling is deliberate rather than incidental: what stresses a plate is its curvature, so a gentle depression of a given depth is genuinely less severe than a sharp one. It also explains why this criterion cannot be checked by a tilt sensor at all, because a sensor reports one slope over one base and a curvature needs three elevations.
Put the four rows together and the practical rule falls out. When a tank is being assessed against a settlement figure, the first question is not whether the measurement is accurate enough but which component the figure protects. A number quoted without its component is not a limit, and comparing a shell criterion with a bottom-plate criterion is a category error that will produce an answer wrong by a factor of forty in one direction or the other.
4. The Allowance That Grows With the Square of the Arc
The out-of-plane criterion in API 653 Annex B is the only one in the ladder that is written as a formula rather than a number, and it is the one that changes how a tilt reading has to be reported. The allowance for local distortion of the shell base is 11 times L squared times Y, divided by twice E times H, where L is the arc length between two adjacent survey points, Y is the yield strength of the shell material, E is its Young’s modulus and H is the tank height. The result is capped at 101.6 mm, which is four inches.
Work one example so that the shape of the formula is visible. Take a shell of ordinary structural steel at a yield strength of 250 megapascals and a modulus of 200 gigapascals, standing 15 m tall. The formula then reduces to a single number multiplied by L squared, and the allowance for a 10 m arc comes out at 45.8 mm. Halving the arc to 5 m does not halve the allowance, it divides it by four, to 11.5 mm. The cap of 101.6 mm is reached at an arc of about 14.9 m, which is longer than the spacing the survey rules allow, so in ordinary practice the formula operates below its cap.
| Arc length L | Allowable out-of-plane settlement | The same figure as a slope | The slope as a ratio |
| 2 m | 1.8 mm | 0.053 deg, or 0.92 mm per m | 1 in 1,091 |
| 5 m | 11.5 mm | 0.131 deg, or 2.29 mm per m | 1 in 436 |
| 10 m | 45.8 mm | 0.263 deg, or 4.58 mm per m | 1 in 218 |
| 15 m | 101.6 mm, the cap | 0.388 deg, or 6.77 mm per m | 1 in 148 |
The last two columns are the important ones, because they show that the allowable angle is not a constant. It grows with the arc it is measured over, from 0.053 deg at 2 m to 0.388 deg at 15 m, and it grows for a mechanical reason rather than a procedural one: the allowance rises as L squared while the slope is the allowance divided by L, so the slope rises linearly. There is therefore no single angle limit for a tank shell, and a statement that a tank has tilted by 0.1 deg is not yet a statement about anything. At a 2 m arc that reading is nearly twice the allowance; at a 5 m arc, where the allowance is 0.131 deg, the same reading sits inside it.
This is where a tilt sensor’s mounting base moves out of the installation drawing and into the specification. The sensor reports one slope over the base it is bolted to, so that base is the arc length of the criterion being applied. Bolt the same sensor to a longer plate and the numbers change without anything moving, because the criterion is being evaluated over a different L. The API 650 survey clause has the same property from the other direction: it requires shell elevation measurements at equally spaced points not exceeding 10 m apart, with a minimum of eight points, and the criterion in API 653 that is expressed per 10 m of circumference has taken its denominator from that spacing. The standard sets the spacing and then writes the limit per unit of spacing, so a survey point grid and a sensor base are answering the same question, and both of them have to be stated before a measured tilt can be compared with a limit.
One further consequence is worth stating because it is the reason the API 650 trigger works at all. Because consecutive survey points are at most 10 m apart, the difference between two adjacent markers is already a figure per 10 m, and the reporting trigger is written in exactly that unit. The spacing rule is what makes the trigger checkable by subtraction, with no division and no assumption about the shape of the settlement curve.
5. What the Survey Reports and What a Tilt Sensor Reports
The level run and the tilt sensor are not two implementations of one measurement, and the difference is not precision. They report different orders of the same field: the survey reports the elevation, which is a potential, and the sensor reports the slope, which is its gradient. Almost every disagreement between a survey result and a sensor result on the same tank dissolves once that is said out loud, because the two have to be related by a difference and a division, and any assumption about spacing or about the smoothness of the settlement curve enters at that step.
| Question | A levelling survey | A permanently mounted tilt sensor |
| What it reports | The elevation of each survey point above a benchmark | The slope of the shell at the sensor’s mounting base |
| What has to be done to get a slope | Difference two adjacent elevations and divide by their spacing | Nothing, the reading is already a slope |
| Sampling gap | The interval between surveys, from the monthly visual check to the external inspection | The reporting interval, from a minute upwards |
| Uniform settlement | Measured directly | Invisible, because a rigid body motion does not change a slope |
| Centre-to-edge dishing | Measured inside the tank, with the tank out of service | Invisible, because the sensor sits on the shell base |
| Localised shell distortion | Recovered from the differences between adjacent elevations | Measured directly, with the mounting base as the arc length |
| Reference | A permanent benchmark, unaffected by the tank’s own settlement | The sensor’s own zero, which has to be logged on day one |
The API 650 survey procedure is worth reading in full before a monitoring scheme is designed around it, because its requirements are a checklist of what a permanent installation has to reproduce. Shell elevation measurements are taken at equally spaced intervals around the circumference not exceeding 10 m, with a minimum of eight points, and the observed elevations are referred to a permanent benchmark. The level instrument itself is set up at least one and a half tank diameters away from the tank while readings are taken. That last requirement is not fussiness: a benchmark close to the tank moves with the tank, and an instrument set up close to a settling structure is measuring the settlement of its own tripod as well as the structure’s.
Six sets of readings are required across the test. One before the test begins with the tank empty, three at a quarter, a half and three quarters of test height, one at least 24 hours after the tank reaches maximum test height, and one after the tank has been emptied. Filling may continue while a set is being taken provided the water level does not change by more than 300 mm across the set. The three intermediate sets can be omitted if the purchaser says so. Alongside the perimeter survey, bottom elevations are measured before and after the test on at least four diametrical lines, at intervals of no more than 3 m, with the lines spaced so that the circumferential gap between them also stays under 10 m.
Two things follow for anyone instrumenting a tank. The first is about the baseline. The record produced by that first fill is the reference that every later in-service assessment is measured against, which is why the standard devotes a clause to the sequence and to the hold at maximum height, and a sensor installed after commissioning is joining a record that already has a beginning it cannot see. The second is about what the sensor cannot contribute. Two of the rows above say invisible, and between them they cover the two criteria in the ladder that a tilt channel cannot check at all: the uniform settlement trigger and the centre-to-edge sagging allowance. A tilt channel on the shell base will not see either, and no arrangement of one on the bottom plate will help, because a membrane that sags to the centre is not tilting locally at all.
6. The Trigger Is Not the Limit
The most useful single comparison in the whole subject sits inside the table in section 2 and is easy to miss. The API 650 reporting trigger is a differential settlement of 13 mm over 10 m of circumference, or a uniform settlement of 50 mm. The API 653 out-of-plane structural allowance, computed for the same 10 m arc on a 15 m shell, is 45.8 mm. Both are describing how far the shell base may depart from a plane over the same length of circumference, and the figure that stops a fill is three and a half times tighter than the figure that condemns a tank in service.
That is not an inconsistency, and understanding why removes a real source of confusion. The two numbers belong to different questions. The API 650 trigger is written for a structure whose foundation is still coming to equilibrium under a load it has never carried before, and its purpose is to transfer the decision. The published explanation of it is precise on this point: the trigger is not a pass or fail limit on its own, it is the point at which the manufacturer stops filling on its own judgement and hands the readings to the purchaser, who decides whether the settlement is acceptable, usually against the API 653 criteria. A number that transfers responsibility is not the same kind of number as a number that says the shell is overstressed, and they differ by a factor of three and a half precisely because they are doing different jobs.
The same table contains the opposite pairing, which is more surprising. The API 653 planar tilt limit of 1 in 120 is 0.477 deg, which is six and a half times looser than the 13 mm per 10 m reporting trigger of 0.074 deg. So a tank may lean as a rigid body by more than six times the local differential settlement that would have stopped its first fill, and both figures are correctly called settlement limits. The reason is the one given in section 1: a lean is a rigid body motion and does no work on the shell, while the local differential bends it, so the two figures are not measuring the same thing at all even though a surveyor with a level and a tape would derive them from the same set of readings.
There is a third asymmetry in the table, and it concerns the tightest figure rather than the loosest. The API 625 peripheral limit of 1 part in 1,000 is 1 mm per metre, which corresponds to 10 mm over 10 m of circumference. The first-fill reporting trigger at 13 mm over the same 10 m is only 30 percent above it. So the trigger during commissioning is not a conservative fraction of the acceptance criterion, it is close to it, and a tank that has just passed the trigger has used up most of the allowance that the design-stage criterion would have granted it. Anyone reading a hydrotest settlement record should hold that in mind: passing the trigger means the readings were handed over, not that the foundation has room to spare.
7. The Interior Between Inspections
The reason a permanently installed channel has anything to contribute on a tank is not that it is more accurate than a level. It is that the survey that can see the most is also the survey that happens least often. API 653 sets three levels of inspection, and the intervals are far apart because the three do different jobs.
| Inspection | Who does it | Maximum interval | Does it see the settlement? |
| Routine in service | Owner or operator personnel | One month | Visually only, as foundation condition and obvious distortion in a walkdown |
| Formal external | An authorized inspector | Five years, or a quarter of the remaining shell corrosion life if that is shorter | Yes, and this is where the in-service settlement survey sits |
| Internal | An authorized inspector | First within ten years, then set by the measured bottom corrosion rate up to a ceiling of twenty years, or thirty with a release prevention barrier | Yes, and it is the only check that sees the bottom plate directly |
The asymmetry is stark. The routine check is monthly but visual, and its coverage of settlement is a walkdown looking at foundation condition. The check that takes a settlement survey is a five-yearly event. The check that exposes the bottom plate, and therefore the only one that can find a centre-to-edge dishing or a depression away from the rim, is a ten to thirty year event that requires the tank to be taken out of service, drained, cleaned and gas-freed, with the outage typically running from two to eight weeks. Across the life of a tank the bottom plate is directly measured on a handful of occasions, and the criteria that govern it are values that have to hold throughout those intervals.
This is the honest case for a continuous angle channel on a tank, and it is deliberately narrower than a sales argument would be. The channel does not replace the internal inspection, it cannot see the bottom plate, and it cannot measure the uniform settlement that the 50 mm trigger is written for. What it does is sample the one quantity that loads the shell, namely the local slope of the base, at an interval set by the operator rather than by a regulatory ceiling, and it does so on a tank that is full of product and in service. Between two internal inspections, that is the only evidence that the shell base is or is not changing.
There is a second reason the interval matters, and it is a statement about the record rather than about the hardware. A settlement that is going to breach the 1 part in 1,000 criterion is slow by definition, and a series of points taken five years apart will show whether the total has grown but not whether the rate has. A foundation that has reached equilibrium under the present fill cycle and one that is still creeping at a fraction of its earlier rate can produce two very similar readings, and separating them needs points inside the interval rather than a better instrument at the ends of it.
8. Placement, the Benchmark, and the Term That Decides
A tank monitoring scheme reduces to five decisions, and the order matters because the first three fix what the fourth and fifth mean.
First, name the criterion and its component. Decide whether the scheme is protecting the shell base against local distortion, the bottom plate against dishing, the footing against differential settlement, or the pipework against rigid body movement, and then write down the figure from section 2 that applies, together with the arc length it is measured over. A scheme that has not done this cannot set an alarm level that means anything.
Second, set the arc. If a tilt sensor is being used, the arc is the mounting base, and it has to be chosen against the criterion rather than inherited from the hardware. A long base reports a smaller slope for the same settlement and admits a larger allowance; a short base does the opposite. What the base cannot do is recover a curvature, because a slope is one number over one length and a curvature needs three elevations.
Third, fix the reference. The API 650 requirement is a permanent benchmark unaffected by the tank’s own settlement, and the level instrument at least one and a half tank diameters away. A tilt sensor has no external benchmark and does not need one, but it does have a zero, and the zero has to be recorded with the conditions it was taken under. The record that a tank assessment needs is not the current angle but the change of angle since a known state, so the day-one entry should carry the fill level, the product, the ambient temperature and the sensor’s own internal temperature and supply voltage.
Fourth, examine what the published terms actually amount to against the criterion, because on a tank the ranking of the terms is not the ranking a buyer expects. Take the tightest criterion in the ladder, the 1 part in 1,000 that converts to 0.057 deg, and compare the three obvious hardware terms against it.
| Term in the datasheet | Published value | Over a 40 degree Celsius swing | Against the tightest criterion of 0.057 deg |
| Resolution | 0.001 deg | Not temperature dependent | 1.7 percent of the criterion, so the criterion is 57 times this term |
| Accuracy | 0.005 deg over minus 15 to plus 15 degrees Celsius | Not published as a temperature-scaled figure in this form | 8.7 percent of the criterion, so the criterion is about 11 times this term |
| Zero point temperature drift | 0.002 deg per degree Celsius on one wireless tiltmeter, and 0.001 deg per degree Celsius with a published maximum of 0.065 deg over the working range on another | 0.080 deg at 0.002 deg per degree Celsius | 140 percent of the criterion at that swing, and the other model’s published maximum alone is 113 percent of it |
The ordering is the payload. The resolution, which is the figure every specification sheet leads with, is 1.7 percent of the tightest tank criterion and cannot decide anything. The accuracy is 8.7 percent of it. The zero point temperature drift, which is usually the last line in the table and is often read as a second-order term, is 140 percent of the criterion over a 40 degree Celsius swing, and on the second of the two models the published maximum drift over its whole working range is 113 percent of the criterion on its own. A shell that sees 40 degrees Celsius of variation across a year is an ordinary case, and the drift scales with whatever swing the site actually has.
The consequence is procedural rather than technical. Because the drift is a slow function of temperature, it does not produce alarm events; it moves the baseline. A tank monitored over a year without a temperature channel will show a tilt that is largest in summer and smallest in winter, and the part of it that is the sensor can be removed by regressing against temperature, while the part that is the shell is genuinely thermal. Removing both is why a tank scheme wants a temperature record whether or not a temperature sensor is part of the product.
Fifth, set the alarm from the criterion and never leave it at a default. The two wireless tiltmeters used in this section both publish a default alarm angle of 3 deg. Against the ladder in section 2 that default is 6 times the loosest limit that protects the shell, 24 times the PIP STE03020 circumferential figure, and 52 times the tightest, so an installation that keeps the default will never alarm on a settlement problem at all. The alarm channel is a separate specification from the measurement channel and should be read as one: the same datasheets publish an alarm accuracy that widens from about 0.01 deg near level to 0.3 deg at the edge of the range, and that upper figure is about two thirds of the loosest tank criterion. The alarm angle has to sit above the structure’s own daily thermal movement and inside the criterion, and that window is narrow enough that both numbers need writing down.
9. The Tank Nobody Can Put a Survey Crew On
There is one class of tank on which the choice between a survey and a permanent channel is not a matter of convenience. Where the stored product is flammable, anything permanently installed in the area has to be certified for the atmosphere it stands in, so a wireless channel is an option only if certified equipment exists for it. The manual alternative is not obviously easier, because a survey that can see the bottom plate means taking the tank out of service and a survey of the shell from the outside is a permit-controlled activity on a tank in that service. The certification therefore becomes a condition of the monitoring scheme rather than a feature of it.
ZC Sensor’s own record in this area is a tilt sensor installed on a natural gas storage tank to monitor the tank’s attitude, with the accompanying note that any electrical equipment used in an atmosphere containing combustible gas requires certification. The unit supplied for that application was an intrinsically safe dual-axis high-precision tilt sensor, and the same product family is described for use in mines and on gasoline tanks. The relevant point for a specification is not the certificate number but the direction of the argument: on a tank of that kind the question is not whether an angle channel would be useful but whether the equipment can be installed at all, and the answer has to be established before the monitoring scheme is designed.
For a tank of ordinary construction, the two wireless tiltmeters quoted in section 8 are the relevant published cases: an NB-IoT tiltmeter with a 0.001 deg resolution and a plus or minus 0.005 deg accuracy over minus 15 to plus 15 degrees Celsius, with a zero point deviation of 0.05 deg and a working range of minus 40 to plus 85 degrees Celsius at IP67; and a dual-axis wireless tiltmeter with a plus or minus 30 degree range, a 0.001 deg resolution and an accuracy of 0.005 to 0.01 deg. The published stocking temperature for the battery in the first of those is plus 25 to plus 30 degrees Celsius, which is a separate point about where the equipment is kept rather than about where it is installed.
The boundary that this article has to state plainly is the one established in section 5. An angle channel on the shell base reports the local slope, which is the quantity that loads the shell, and it reports it continuously on a tank in service. It does not measure uniform settlement, it does not measure the centre-to-edge dishing that governs the bottom plate, and it does not measure a curvature. The judgement that a measured settlement is acceptable belongs to the tank’s inspector and to the geotechnical engineer who assesses the foundation, working from the criteria in section 2 and the record from section 5. A monitoring scheme’s job is to make sure that judgement has a current record to work from between the surveys that the standard requires.
Frequently Asked Questions
- What is the allowable settlement of a storage tank? There is no single figure, and quoting one without naming the component is the source of most confusion on the subject. The published limits run from about 1 part in 1,000 for the differential settlement around the periphery of a new tank, which is 0.057 deg, to about 1 part in 25 for the sag of the bottom plate, which is 2.29 deg. The shell, the bottom plate and the ring footing each have their own limit because each is damaged by a different aspect of the same ground movement, so the answer always begins with the question of which component is being assessed.
- Can a tilt sensor replace the tank settlement survey? No, and the two are not interchangeable in either direction. A survey measures elevation and derives a slope by differencing two readings, so it can produce uniform settlement and the centre-to-edge profile of the bottom plate, which a tilt sensor cannot see at all. A tilt sensor measures the local slope directly and can do so continuously on a tank in service, which a survey cannot. What the sensor changes is the interval between usable records of the one quantity that loads the shell, not the survey’s role in the assessment.
- My tank is leaning, but the shell shows no distortion. Is that a problem? It is a serviceability problem rather than a shell stress problem, and it needs to be assessed against the planar tilt limit rather than the out-of-plane one. A lean is a rigid body motion, so it puts no additional stress into the shell, but API 653 lists the consequences of excessive planar tilt as floating roof seal malfunction, floating roof binding, problems with connecting pipes, drainage problems on the pad, and shell buckling. On a tank with a floating roof the rim gap is the governing detail, and the check is usually the one that decides how much lean can be tolerated.
- What alarm level should I set on a tank tilt sensor? It has to come from the criterion that the scheme is protecting, converted into the arc the sensor’s mounting base covers, and then raised above the structure’s own daily thermal movement. The default alarm angle published on the two wireless tiltmeters quoted above is 3 deg, which is 6 times the loosest published limit that protects a shell and 52 times the tightest, so a tank left on the default will not alarm on settlement. The other figure that matters is the zero point temperature drift, which on one of those models reaches 0.080 deg over a 40 degree Celsius swing and is therefore larger than the tightest criterion itself.
TAG: