Building on a sloping site: what actually makes it expensive

The sloping lot is nearly always the cheapest one on the street, and nearly always has the best view. Both things are true and the two are connected: it sells cheap because building there costs more, and the buyer usually finds out how much more on the day the construction budget arrives.

What makes a hillside house expensive is not the architecture. It is the ground, and it comes down to three items that were not in the figure you were shown when you bought the lot: the geotechnical study, the stabilisation works, and the retaining structures with their drainage. None of the three is optional, and all three are named outright in the building code.

The soil study, and exactly how much of it

"Get a soil study" is the most repeated and least useful piece of advice, because nobody says how big a study. Título H of NSR-10, the part of the code that deals with geotechnical studies, does say, and it says it by size of project.

I quote the whole sentence deliberately, because the second half falls off every time somebody summarises it. On urban and suburban land — two of the three classes Colombian planning law uses, the third being rural — the study is compulsory for any use, houses included. Beyond that, the duty stated in this clause reaches use groups II, III and IV — buildings of special occupancy, community service and essential facilities — and a single-family house is group I. That does not make the study dispensable: H.5.2.1 demands it again the moment there is a slope, and no serious structural designer sizes a foundation without one.

The size of the study follows from classifying the project. The code calls the building a "construction unit" and sorts it into four categories by the total number of levels and the maximum service loads on columns, always taking the less favourable of the two.

A two-storey house is low category: three boreholes and six metres. That is the floor, not the ceiling. The same title makes clear that meeting the minimum does not release the engineer from drilling whatever additional boreholes professional judgement calls for, and that if they are not drilled the recommendation must be recorded in writing in the report.

And there is a threshold worth holding in mind when buying, because it decides how complex the study becomes: if the slope of the ground exceeds 10 per cent, the code treats topographic effects as relevant and calls for dynamic response analysis using two- or three-dimensional numerical models. Ten per cent is a slope you can barely see by eye.

Cut and fill: why the fill is the enemy

Putting a house on a hillside means making a platform, and a platform is made by taking earth off one side and putting it on the other. The excavated side is the cut; the built-up side is the fill. Once the work is finished and the slab is poured, the two halves look exactly alike. They do not behave alike.

The reason fits in one sentence, and it sits in the foundations chapter of the same title.

Read it through, because the two halves say different things. Fill is not forbidden: loose fill is. An engineered fill — selected material, free of organic matter, placed in layers and compacted under control — is a legitimate bearing stratum. The heap of earth that came out of the cut, pushed around with the excavator and levelled by eye, is not. And to the naked eye they are identical.

The code does not leave it to the eye. For cohesive fills it requires layer compaction to be controlled with the standard Proctor test, watching layer thickness, water content and the rate of placement in height; for non-cohesive ones, control rests on relative density. A fill without compaction testing is not a fill: it is earth that has been put there.

The retaining wall, and what actually fails

A retaining wall holds earth back, and the earth it holds can beat it in six different ways. The code lists them as ultimate limit states: structural rupture, deformation, overturning, bearing capacity failure, loss of support through erosion of the ground, horizontal sliding of the base and, where relevant, general instability of the slope the wall is founded on.

Of those six, the one you see on site almost always starts with water. A wall is designed for earth pressure; if water builds up behind it, water pressure has to be added, and that pressure rises fast. Which is why the code uses a word here that it hardly uses anywhere else.

Always. And the same clause adds the warning that turns drainage into a piece of design rather than a pipe: the filters must be designed so as not to carry fines out of the fill and so as to conduct the water without generating significant pressures, and even with a drainage system, the effect of seepage forces on the pressure can be significant.

How much margin the code demands

There is a question an owner can ask that separates a serious report from a box-ticking one: what factor of safety did the slope come out with. The code sets the minimums and here they are.

A slope with a factor of 1.05 under seismic conditions is not "bad": it is at the minimum the code allows, which is a different thing from having room to spare. Knowing where on that band your site came out is information you paid for and are entitled to have explained.

Access, and the water coming down

Two matters that weigh more on a hillside than on any other lot and that the seismic code does not settle.

The first is vehicle access. A long steep ramp is a work in its own right: walls, drainage, paving with a non-slip finish and, at times, more square metres of structure than the house itself. And it is the first thing to resolve, because without access not even the mixer gets in.

The second is runoff. On a slope, rainwater does not stay: it goes down, and it goes down wherever you let it. The platform you built is a new impermeable surface that concentrates flow, and if that flow ends up discharging onto the fill batter or onto the neighbour below, the problem is yours twice over, technically and legally. A cutoff ditch above the cut, collection on the platform and a defined discharge point are budget items, not details.

The items that are never in the first estimate

1. 1The definitive geotechnical study, with its number of boreholes and their depth counted from the original ground rather than from the finished floor.
2. 2The slope stability analysis and the design of the stabilisation works, which H.5.2.1 requires and which is a separate job from the foundation study.
3. 3The retaining walls with their filter and drain, budgeted by square metre of wall and not by linear metre of facade.
4. 4Earthworks and the removal of surplus material, which on a slope almost never balances out within the lot itself.
5. 5The vehicle access in full, including its retaining structures and its drainage.
6. 6Runoff management, with a cutoff ditch and a defined discharge point.
7. 7The geotechnical engineer on site, which the code requires for the medium, high and special categories, and which on a hillside is worth hiring even when the house is low category.

None of these seven appears in the cost per square metre that circulates, because that figure belongs to the house and these belong to the ground. If you are putting the budget together, the conversation continues in what a country house costs. If the land is rural, the permit runs on its own calendar and that is in why a rural building permit takes twice as long. And the water coming in, as opposed to the water coming down, is a whole other conversation: what permit it needs.

A hillside handled well gives you the best house on the street. Handled badly it gives you the one that cracks in its third rainy season. The difference between the two is decided before the plan is drawn, and it costs less than repairing it does.