The Tank Does Not Try to Float When It Is Full. It Tries When It Is Empty.

A buried tank on a site where the water table stands about two metres below grade will try to float. Not when it is full, which is what most people picture when somebody says the word structural. It tries when it is empty.

That sentence already exists on this blog. It is one line inside a longer note on whether the tanks go above ground or below it, where I wrote that a tank standing empty in a wet season can try to float, called it a design case rather than a surprise, and moved on. This is the note that line was promising. It is not about where to put the tank, and it is not about what the layout does to the civil bill or the order the thing gets poured in. It is about the single load case that sized the base slab on a job earlier this year, and about why that load case is the one nobody in the room asks about.

The case everybody pictures is not the case that governs

Ask somebody what the worst day is for a buried sewage tank and they will describe a full one. Water inside pushing out, soil outside pushing in, a lorry parked on the cover slab. That is a real set of cases and it sizes the walls and the cross walls, and it is run at more than one loading rather than one, and it is the picture that goes with the drawing in a sizing and specification note.

Now empty the tank and leave the groundwater where it is. The hole around the structure is full of water up to the level of the table. The tank is a sealed box sitting in that water with nothing inside it but air. It is a boat. The water pushes up on the underside of the base slab across the whole plan area, and the only thing holding the tank in the ground is what it weighs.

On the job in question the floor of the deepest cell sat about five metres below grade and the client told us the table comes up to about two metres below grade at times. That is three metres of water standing above the underside of the slab, across the whole footprint of an eight-cell box roughly eleven metres by six. The uplift is that head multiplied by that plan area, and water weighs about a tonne per cubic metre, so the number gets large quickly and it does not care at all how good the concrete is.

The two numbers, and the ratio between them

The calculation is honestly not complicated. One number is the upward push: the plan area of the base, times the height of water standing above it. The other is everything acting downward: the dead weight of the concrete, plus anything permanently sitting on the structure, plus the weight of soil and water bearing on any part of the slab that sticks out past the walls. You divide the second by the first and you get a factor of safety. Below one, the tank lifts.

On the first pass that ratio came out at 1.11. The tank was heavier than the water it displaced, so it was not going to float. It was heavier by about a tenth, which is not a margin. A tenth is inside the range of things nobody controls on a site: how much of the wall thickness is actually delivered, whether the backfill ends up as compacted soil or as something closer to mud, whether the table comes up higher than the client remembers in a year like the one that brought the Cebu rains we wrote up in August. We want to be clear of a fifth, and different engineers will tell you different numbers for that; I am telling you ours rather than quoting a rule at you.

So the design question stopped being how thick is the slab and became what else can be made to hold this box down.

The fix was geometry, not thickness

The instinct is to pour a heavier slab. It works, in the sense that arithmetic works, and it is the most expensive way to buy a given number of tonnes, because every extra tonne of concrete is also a tonne somebody mixes, carries down into a hole and finishes. Thickness is a terrible lever for buoyancy.

What we did instead was extend the base slab 750 millimetres past the wall line, all the way round, as a flat collar. The slab is already being poured, the formwork edge simply moves out, and the reinforcement runs on. What that collar buys is not its own weight, which is small. It is that the soil and the water sitting above the collar are now standing on the structure. They were always there; the slab was just not reaching under them.

That one change added roughly a third to the hold-down side of the equation and moved the factor from 1.11 to 1.46. No extra thickness, no extra chamber, no piles, and nothing that anybody has to maintain for the next twenty years, which is the after-sales test we put every specification through. It is the cheaper half of the same trade we make on anything that has to be paid for every month rather than once: spend at the pour, not forever. It is also the same instinct that made us build the clarifier hopper out of mass concrete on the flat slab rather than cast a shape below the foundation line: solve it with the geometry you are already paying for.

There was one more thing in our favour on that site. The client wanted a two-metre clear setback on all sides of the tank, for his own construction reasons. That setback is what makes a 750-millimetre toe and an externally applied waterproofing coat buildable at all. On a tank pushed up against a boundary wall you may have the arithmetic and no room to execute it, which is a constraint that belongs in the risk and assumptions section of the quotation and not in a conversation after the dig has started.

The empty tank is not a hypothetical

The reason this case is easy to wave away is that nobody plans to leave a sewage tank empty. But the tank is empty more often than the picture in anybody's head suggests.

It is empty from the day the cover slab cures until the day somebody starts putting water in it, and on a normal build programme that gap is weeks, sometimes months, because the plant is finished before the building that feeds it is occupied, and on a building let in phases the tank is full long before the tenancies are. If those weeks land in a wet season, the tank is sitting in the worst version of its own worst load case with nobody watching it. It is empty again whenever a cell is drained for an internal inspection, for repairs to a wall coating, or for the kind of retrofit work inside an existing tank that we get called to do on a plant being brought back into service. And it is partly empty every time somebody desludges aggressively, which is a real risk on a plant where everything taken out of the plant has to go somewhere and the wasting is run off a vacuum truck calendar instead of a cone reading.

There is an operational answer to all of that, which is to say in the handover documents that the tank is not to be fully drained without the groundwater being checked or a cell being kept full. We write that down. But an instruction that has to be remembered by whoever is on site in year seven is a weak defence, and on a plant that has already been through two or three changes of operator it is close to no defence at all. Design it out and the procedure becomes a courtesy rather than a dependency.

Waterproofing from the inside, and the one exception

The second half of this job ran against normal Philippine practice, so it is worth saying plainly what that practice is. Sewage tanks here are waterproofed from the inside. The coating goes on the wet face, inside the chamber, and it is there to stop the contents getting out into the ground and into whatever the neighbours are drawing from. Nobody applies an external membrane to a buried sewage tank in this country as a matter of course. I have not seen it specified on an ordinary job.

Where the groundwater is high, the argument runs the other way as well. If the table stands above the floor of the tank for part of the year, the ground is trying to push water in, through construction joints, through penetrations, through whatever the waterstop at the kicker did not quite catch. Groundwater leaking in is not a pollution problem but it is a treatment problem and a declared flow problem, because it is hydraulic load arriving at the plant that nobody put in the design and nobody is paying a water bill for. On this job the client asked for waterproofing on both the positive and the negative side himself, which is the first time anybody has asked me for that, and on a site with the table two metres down he is right.

Both sides is not a free upgrade. It needs the working space outside the wall, it needs the external coat to survive backfilling, and it adds a sequence step between the pour and the backfill that has to be inspected before it is buried, because afterwards it is permanently unreachable. That is a civil works scope item with a hold point in it, and if the civil works are being built by somebody other than us it has to be written as an acceptance check rather than an assumption.

The spare depth, and why it is in the file

One number from that analysis is worth explaining because it is the reason a design can absorb bad news. When we ran the uplift case we also ran it for a deeper floor, and the factor stayed above a fifth of margin down to roughly a metre and a half below the floor we had drawn.

That is not a boast. It is the answer to the question every owner eventually asks, which is what happens if the sewer arrives lower than the drawing says. Incoming invert levels move. They move because the building drainage was re-routed, because the 200-metre run to the plant turned out to fall more than anybody surveyed, because the civil contractor hit something. On this job the main arrives shallow and the equalisation tank was sunk and its working level set so the top water sits below the arriving invert, which deleted a separate lift station from the job entirely. I have written that one up separately, because deleting a structure by choosing a level is a decision worth its own note. The point for this one is that knowing how much deeper the floor can go before the structure stops being safe against flotation is the difference between absorbing a level change and redesigning the tank in the middle of a dig.

A design that has no idea where its own limits are is a design that has to be re-done every time the site says something unexpected. That is the same argument as pricing from rates we have actually paid, applied to geometry instead of money.

If the soil above the toe is doing the work, the backfill matters

There is a consequence to solving uplift with a collar that is easy to miss, and it belongs on the drawing rather than in somebody's head. If part of the hold-down comes from the material standing on the toe, then the material standing on the toe is now structural. Backfill that was going to be whatever came out of the hole, tipped back in loose, is no longer a neutral choice.

So the note on the drawing says what goes back against the wall and over the collar, and in what layers, and that it is placed and compacted rather than dumped. It also says the external coating is to be inspected before any of that happens, because there is exactly one moment when it is still reachable. This is the same class of item as the chamber order that got settled by where the incoming invert landed: a decision whose cost is nothing at the right time and large at any other.

It also changes what a handover is. A buried tank hides almost everything about how it was built, which is why we ask for the water tightness test to be witnessed and why the acceptance checks are written down before commissioning starts rather than negotiated afterwards. Nobody is excavating a toe in year three to find out whether the compaction was done.

What an owner can ask before the slab is poured

This is the part that is cheap to check and impossible to fix afterwards. If you are looking at a set of drawings for a buried plant, five questions get you most of the way.

• Which load case governed the base slab. If the answer is the full tank, the empty case has not been looked at.
• What water table level was assumed, and where that number came from. A borehole log, a neighbour's well, or somebody's memory of a bad year are three very different things, and an assumption is fine as long as it is written as one.
• Does the base slab extend past the wall line, and by how much. If it does not, the only thing resisting uplift is the weight of the structure itself.
• What the structure weighs against what it displaces, as a ratio, and what ratio the engineer was designing to. You do not need to check the arithmetic. You need to see that somebody did it.
• What happens the first time a cell is drained. If the answer is an instruction in a manual, ask who will be reading that manual in ten years.

None of those five are questions about concrete strength, which is the thing most people ask about. Three thousand psi is more than enough for these tanks. Flotation is not a materials problem, and it is not something a stronger mix or a clause copied into a specification from somewhere else can buy you out of.

Where it does not apply

Most of this disappears on a dry site. Three kilometres inland with the table ten metres down, the empty case never governs and the base slab is sized by bending, as it should be. If the plant goes above ground the case does not exist at all, which is one more decision that is taken at design stage rather than discovered later, alongside the depth you do not have to buy for the clarifier.

The toe is also not the only route, and on a larger or deeper structure it stops being the best one. There are others: a separate ballast slab, tension piles anchored into competent ground below, or a pressure relief valve in the floor that lets groundwater into the tank rather than resisting it. We did not use a relief valve and would not here, because it means deliberately admitting groundwater into a chamber whose contents you then have to treat, and we would rather carry the structure than carry the flow. On a very deep or very large tank that trade can reverse, and at that point it is a structural engineering exercise in its own right rather than something settled in the same week as the choice of treatment technology.

The waterproofing half travels less well than the structural half. Inside-only is the practice here and it is the right default for most sites. Both sides is the exception, it is bought for a reason, and it is worth nothing if it is applied badly and then backfilled over. And a plastic or glass-reinforced tank sitting in the same hole has the same problem in a sharper form, because it weighs almost nothing. It gets held down by its surround, which is a different calculation from the one above and not one to borrow.

What generalises is only the habit. On any buried tank, ask what the empty case says before you ask anything about the full one. The full tank is the picture. The empty tank is the one that lifts.

If you are looking at drawings for a buried plant and nobody has shown you an uplift check, we are happy to look. You can see the work we have done, read how we approach wastewater treatment and what the compliance limits actually require, size a plant yourself on the price estimator or the design generator, look up the parameters that matter in your industry, or send us the site through a proposal request or the get an STP page. We work across the country as a wastewater treatment company and an STP contractor, we are based in Cebu, we take on operation and maintenance, and our consultancy work is set out separately along with where the free advice ends and the paid work begins. Or just book a free 15-minute consultation and bring the section drawing.